Reports & Briefs

Platforms & Publishers

Reports & Briefs

Tow/Knight Projects

Sensors and Journalism

Project Leader:

Acknowledgements The hardworking journalists, scientists, activists, and researchers featured in the case studies gave a lot of their time to talk with the Tow Center, review drafts, and clarify fine points. In some cases they went back to years-old work to dig out important context for current trends. Really, any valuable analysis in this report is built on their work, so without their efforts, this would all be imaginary. They are: Dina Cappiello, Thomas Stock, Jeffrey Warren, Liz Barry, Shannon Dosemagen, Ben Gamari, Don Blair, Mat Lippincott, John Feighery, Alison Young, Blake Morrison, Sally Kestin, John Maines, Guan Yang, John Keefe, Josh Davis, James Jensen, Brian Boyer, Brendan Schulman, David Fallis, Matt Waite, and Mickey Osterreicher. The authors of the legal and ethics collection enthusiastically powered up their experience and talents to shine a light on this field, despite our (then) lack of definition and way too much hand-waving from me. Then, having figured it all out, they endured our nitpicking through more rounds of editing than I’d signed them up for. The result is a set of essays laying out a legal and ethical base that the industry can refer to in the years ahead. Their biographies are on page 1. On behalf of everyone they will help keep out of trouble, “Nice one.” At the Tow Center, Director Emily Bell and Research Director Taylor Owen gave me great direction and huge amounts of freedom and trust. They assigned generous funding from The Tow Foundation and the John S. and James L. Knight Foundation so I could work with wonderful people, and gave useful feedback at critical points through the drafting process. This was a serious opportunity, for which I am extremely grateful. Back at the very start of my interest in this subject, Professor Mark Hansen generously showed me all the work that had come before, and introduced me to ideas, analysis, and so many of the brilliant people he attracts. Charles Berret stepped in midway to tell a masterful story of sensor history. His scholarship dazzles. Lauren Mack kept everything on rails, explained how the place worked, got people paid, and this report printed. Thanks. Finally, my deepest appreciation goes to Abigail Ronck, who edited this report like a champion. I sent her drafts full of messy syntax, terrible spelling mistakes, punctuation errors, and flabby paragraphs. She uploaded copy with clean, sparkling sentences, told me when my thinking had holes, and filled in more of those gaps than I’d prefer to admit. As the word count grew she remained encouraging, committed and, perhaps most importantly, interested. Thanks. A lot. —Fergus Pitt, May 2014


Introduction & Executive Summary


This report you have opened is not monolithic. You can match its sections to your needs. The first section introduces our topic. It starts by describing the landscape where sensors and journalism combine, and continues on to define necessary terms for understanding this area of research. Reporters are using sensors in an era when the rapid development of technology is moving data into the mainstream of journalism. The increasing ubiquity of sensors, their increasing capability and accessibility are on the supply side, while investigative reporters, computer aided reporters and journalist/technologists are on the demand side. We are including drones in the field of sensing, partially because of the amount of attention they’re currently receiving, and partially because of their potential to extend human sight far beyond our bodily bounds. While recent commentaries about journalistic sensing have focused just on sensors that journalists have built themselves (or commissioned), our definition also includes journalistic uses of data from sensor systems that are not controlled by the reporters themselves. We have excluded opinion polling, information gathered by humans’ five senses, and data produced by monitoring computer processes like bit-torrent networks. That said, our description should not be used to separate sensor-based journalism from other reporting processes. The intellectual tools we discuss may be useful for many data-intensive projects, and sensor reporting needs to be integrated with traditional forms.

Then, scholar Charles Berret has written a chapter on sensor history, charting humanity’s efforts to extend the reach of our five natural senses. It starts with the scales unearthed by archeologists, the Neolithic markers like Stonehenge, and the agricultural tools from the Nile region. Berret notes that, in the 1500’s, the astronomer Tycho Brahe built a data network using the post, which compiled sensor readings to draw the most accurate and comprehensive star maps of his time. Cameras and sound sensors came in the nineteenth century, a moment ’when mechanical sensors were first treated with greater credibility than the human observer.’ The history outlined here only goes as far as the first half of the twentieth century, but within our time period it covers early medical sensors in the form of René Laennec’s stethoscope and Willem Einthoven’s electrocardiogram, and meteorologists’ use of doppler radar. The introduction finishes by outlining the characteristics of sensors that make them useful—or not—and helping readers identify what elements of the world can be sensed. The second section, containing case studies, examines seven projects that used sensors for journalism. Each study includes the story of what happened and then offers analysis in which we identify its distinctive or noteworthy elements, as well as the lessons journalists may take from the projects. The case studies start to show distinct types of sensor uses that suit different journalistic goals. The first type is when investigative reporters (environmental reporters in these two examples) design a sensing process to collect data with the intent of testing a hypothesis. They used relatively mature professional equipment and consulted with experts. They had justifiable confidence in their data, even though their processes were quite different from how scientists work when intending to publish in a peer-reviewed journal, or when doing work on behalf of regulators.

Another type of sensor use by journalists is accessing data from municipal sensor systems. The Sun Sentinel won a Pulitzer Prize by using tollgate data to investigate widespread speeding by off duty Florida police. The Washington Post published an extensive explanatory feature based on data from a network of microphones installed by the city law enforcement. A separate type of journalistic sensing involves DIY hardware development. At the moment, these projects value participation and informal science education. At the moment the equipment they use is unlikely to produce data that can be heavily relied upon in legal or health settings. However, the makers in this part of the field see great long-term potential, inspired by open-source software, a phenomenon that has returned great value to newsrooms. In case studies, we have also added analysis of the U.S. drone journalism industry, as it stands right now. At the high end, a small number of organizations are using footage shot by specialist pilots of professional cinematography drones. In the middle, enterprising news industry employees are experimenting with pro-am equipment costing hundreds, not thousands of dollars. Mainstream media organizations are also sourcing drone footage shot by hobbyists. All of this activity is proceeding despite a rapidly changing, highly contested regulatory environment. For the next section, about laws and ethics for reporting with sensors, we recruited 12 experts to write a chapter each. They applied their considerable knowledge and ability from professions in law, technology, ethics, academic research, and the sciences. In the individual essays, each helps identify and navigate the key issues that arise when their field intersects with sensors and journalism. The authors who address the privacy and surveillance issues write that these are early days for the field. The courts have thus far dealt with consent to record, defamation and false light in the context of cameras and microphones. The potential for journalists to break those laws using different types of sensors certainly exists, and if legal claims are made courts will

likely consider the ethical standards that emerge in these next few years of sensor reporting. The field is emerging even as relationship between newsrooms and their audiences transforms. Our authors suggest that journalists should involve their communities as they negotiate the tricky questions of who owns and controls personal data from sensors. Newsroom managers who have staff making or acquiring hardware should also acquaint themselves with the basics of open-source licensing. Often, journalists who design their own sensing systems will lean towards sharing their work under open-source principles, but this may involve legal liability if hardware goes wrong and causes physical damage. The risks are avoidable, however, as the article by Diana Cooper makes clear. Still in the realm of legal issues for hardware makers, this current phase of rapid, widespread DIY development is moving a lot faster than The Federal Communications Commission. The FCC requires that any electronic device that might emit radio interference be tested and approved before marketing. However, that regime did not consider many conceivable journalistic uses of custom sensors produced in small batches. If and when sensing, in particular drone use, becomes a widespread journalistic practice, human error is likely. Serious mistakes will attract negligence claims, following in a tradition codified by The Digest of Justinian in the mid-sixth century. It contained a section on ’Those Who Pour or Throw Things Out of Buildings’. The laws extended to falling things, as well. Despite this history, the novelty of drone journalism will make insurance tricky and expensive until the industry has more data on which to model risk profiles. The last group within the legal and ethical section concerns truth and accuracy. Sensors may seduce journalists into thinking their output is objective and free from the errors inherent in human testimony. That is a risky belief. We have drawn on the expertise of the EPA to show how reporters might design a sensor-based data collection process to improve their accuracy. For journalists, the concept of ground-truthing—supplementing sensor

information with human input—will be valuable. It should help introduce nuance, guard against mistakes and treat fairly the people at the heart of our stories. For the final section, we have distilled this report into a set of recommendations, including groups of strategic moves, good work practices, and efforts the industry may collectively consider. Strategic Recommendations

  • Identify and cultivate sensor sources for the beats you’ve prioritized.
  • Put a watching brief on open source sensing systems.
  • News nerds should do hardware too. Work Practice Recommendations
  • Before sensing, articulate your hypothesis.
  • Work with experts on complex stories.
  • Understand the entire pipeline for your story’s sensor data.
  • Combine sensing with traditional reporting. Recommendations for the industry, collectively
  • Journalists have an opportunity and a responsibility to report on sensor systems.
  • Advocate for access to data from publicly funded sensor systems. However, even though we have documented significant amounts of journalistic sensing here, we hope that this report will need updates as newsrooms keep combining their reporters’ ideas with new sensing opportunities.


The First Section: A Framework

The Landscape for Sensors and Journalism

Throughout 2013 and 2014, whenever the Tow Center gathered people together to work on the topic of sensor journalism, we needed to set the scene. When it comes to labels, “sensor journalism” isn’t even as understood as “data journalism.” So, although the last thing journalism needs is a new term to define a fragment of its practice, we must draw some boundaries and describe the landscape to bring readers of this report into a common language. Nonetheless, we reserve the right to retreat or advance from this ground, or even open the borders, as seems necessary in the rapidly evolving world of journalism circa 2014. By erecting a frame of reference for journalistic sensing we hope to help readers efficiently understand which intellectual tools they can apply to their own sensing and sensor-based work by others. However, by describing this field discretely we do not mean to advocate that it be practiced discretely from other types of reporting; quite the opposite. Indeed, the legal and ethical sections and most of the case studies that follow demonstrate the value in combining sensor-based reporting with other journalistic tools, including personal interviews and shoe-leather reporting, so that the data can be incorporated with context, narrative and emotion.

King Data and the Wide Angle for Journalism First, we offer some observations about why it might be worth focusing on sensors in combination with journalism. This is the context for this report; a suggestion of why there is value in this particular vein of research. Sensors are a way of collecting information about the world. Journalists trade in acquiring information, analyzing it, organizing it, and distributing it. That alone suggests a natural fit. However, beyond that, journalists are currently paying special attention to data that can be easily parsed by a computer. Many readers of this report will be familiar with the impact of data-specialist teams put together by the top tier of American and international news companies, whether in longestablished newsrooms like The New York Times, the LA Times, the Guardian, and the Washington Post, or ambitious newcomers like FiveThirtyEight, Buzzfeed, and Vox.1 Journalism schools have recruited professors who teach data analytics, principles, and presentations. Hundreds of Master of Science students entering Columbia University’s Graduate School of Journalism in 2013 took at least three seminars about data practices. Many go on to learn Python and R programming languages in greater depth. The primary U.S. conference for data journalism, NICAR, went from a conference of a few hundred people in 2010 to one with a thousand attendees in 2014. The rise of data journalism has coincided with an age in which technology is becoming cheaper, more capable, and more widespread. Many observers would suggest a causal relationship: When computers permeated more homes and schools, more children learned programming skills. More systems and data about the world have been digitized. There are more stories to be found in databases and more journalists working in the profession with the interest and skills to find them. Sensors Everywhere The hardware components of technology, including sensors, are also cheaper and more ubiquitous. While the following examples of sensors don’t encompass the breadth of technology observed in this report, they can be a useful illustration. Cellphones include cameras, accelerometers, and GPS sensors, microphones and radio frequency receivers. Those components now cost a few dollars at wholesale. Young children (or at least their parents) can spend $36 to buy a pair of sneakers that sense movement and trigger small lights in the soles. Sensors are baked into civic processes like traffic control, and industrial processes like stock control. A private company called Planet Labs has put 28 toaster-sized satellites into orbit, designed to operate as a flock of cameras pointed toward the earth. It’s aiming to have an array of 100 craft in space by March of 2015. We are living in a sensed world. Aside from the sensors incorporated into finished products for consumers, governments, and private enterprise, sensors comprise a key component class used by the “maker movement.” That ecosystem encompasses electronics prototyping platforms such as Arduino and Raspberry Pie, DIY electronics retailers Sparkfun and Adafruit, and, if interpreted widely, also includes KickStarter’s crowd-sourced product development and the consumer-side of 3D printing. Taken together with the spread in programming skills, it is fair to say that DIY hardware development is flourishing. In early 2013 the makers of the popular electronic prototyping platform, Arduino, said there were 700,000 boards in operation. They estimate that the number doubles if one includes the clones (which are legal, given their open-source license).

So for journalism, there is a special symmetry of demand and supply. Behind the computer-aided reporters and data savvy journalists have come a generation of programmer-journalists. They all compete for fresh data to include in their stories. Another artifact of the digital-first era is the development of news apps and interactive news graphics, in which users’ ability to explore data requires that information be available in granular form, not just as a summary paragraph or a static graphic (although those are also both perfectly justifiable outcomes of a data-reporting process). As the case studies in this report explore, sensors can produce the data demanded by computer-aided journalistic processes. However, beyond simply satisfying the existing demand for data, particular characteristics of modern sensors and their accompanying technologies produce opportunities for new reporting processes. The low cost of some sensor types enables experimentation and new modes of use. In the case studies that follow, we see examples of cameras being practically disposable, or at least sacrificial. Environmental sensors might be widely deployed and left always-on. Cameras can be used, not for the scenes they depict, but as a source for a pixel-by-pixel computational analysis. The sound waves picked up by a microphone have been parsed for the characteristic signature of a newsworthy event. Why Would Journalists Want to Sense? When the Tow Center ran a workshop (in June of 2013), hosting a range of journalists, researchers, and technologists we asked participants about their hopes and ideas for sensors in journalism. A couple of themes emerged: The first was simply a desire for more data to use in their reporting. Especially in the environmental sphere, our participants felt there was a deficit in the data being provided by official sources. They wanted data with better spatial coverage—often targeted where they expected to find problems but couldn’t be sure. Pollution monitoring near industrial facilities was the most common example. That desirability of having data about more places can logi cally be extended to increased temporal resolution. Instead of taking a sample once a month or once a week, some journalists want to monitor aspects of the world all the time.2 In the case studies, you will find examples of sensors used to collect with greater spatial and temporal density—not just on environmental topics. However, as well as getting data with more resolution, the case studies show imaginative applications of sensors providing different types of information, especially around the location of people over time. We heard another potential benefit of using sensors in journalism: to take human observations and impressions and make them specific, so that they might be used for comparisons. Often that meant quantifying an observation: The amount of a chemical in the air matters if it is to be compared to a known health risk factor; the speed of a car matters if it is to be compared to a law. But journalists also want to make comparisons across time and space; does one neighborhood in Washington have more gunshots than the next? Has the number gone up or down over the last year? A sensor can record an aspect of the world so that it can be specified and transparently communicated. These aspects, we believe, are the context for our research into sensors and journalism. We are in an era in which reporters are hungry for data, and increasingly expert in using it; in an age when sensing technology is developing radically and permeating every aspect of modern life. Those trends, taken together in journalism, have produced new demands that sensors might meet, opportunities that might be exploited, and benefits that might be realized.

Journalistic Sensing in This Report The case studies in this report document journalistic projects using sensors that clearly fit within these trends. An early classic in participatory electronic sensing was The Cicada Tracker, which saw WNYC listeners build Arduino-based sensors to contribute readings of temperature readings in their local environment. USA Today’s multi-year effort to take almost a thousand soil samples became the Ghost Factories project, and picked up numerous investigative journalism awards. Nonetheless, through our discussions and work over the last year, we’ve found ourselves pausing on specific journalistic projects and asking whether or not they fit into the ’sensor journalism’ field. Again, we have no desire to make delineations for their own sake, but simply to demonstrate how this framework applies. We’ve worked through a few examples to help readers understand why they are reading various chapters that follow. Drones seem to be worth including in this report. To date, most journalistic uses of drones have been for collecting photos and videos (although other industries also leverage drones’ ability to collect 3D landscape data and environmental data). While cameras have been used in journalism since they were invented, the qualifying aspect for drones, we believe, is that they leverage the radical advances in camera miniaturization. Insofar as drones carry sensors and thereby extend the reach of reporters to observe and record the world, they fit into this field. Also, through the second half of 2013 and the start of 2014, civic uses of drones—including journalism— have become the topic of increasing research, mainstream media attention and regulatory action. Many newsrooms we spoke to have started planning for drone use. We think including drones in this report makes it more relevant to an urgent conversation in the news industry.

Likewise, we have included journalists’ use of data produced by sensors they didn’t commission or control. There’s a good counterargument to be made that this isn’t distinguishable from data journalism, but this report is not trying to carve out a field of sensor journalism that is apart from data journalism. We refer again to our goals for this report: to help journalists use these sensors as well as they can, and to help them understand data that comes from sensors. While some journalistic uses of sensors do involve building customized sensors, or running their own sensing programs, we see no reason to narrow the discussion to those use cases. Two of the case studies deal with data that was released when journalists asked for data under Freedom of Information Act principles. Although we haven’t included a case study about journalists’ use of remote sensors on satellites, we know of one newsroom investing a lot of time to understand and use that data for its work. In the coming months, readers can expect to see more stories based on innovative uses of remote sensing. Data Collection Beyond our Purview On the other hand, there are some related journalistic practices we’ve left out, even though they may share characteristics with the history, theories, and journalistic uses of sensing. In the process of drafting his paper on the epistemological considerations of sensing, University of Wisconsin assistant professor Lucas Graves wrote a provocation to include polling. Like sensing, polls are a way for journalists to systematically collect information about the world, often in a form ready for computation. Bloomberg and Reuters use consumer and business sentiment polling on a weekly basis. They brand strategically important polls and form partnerships with polling organizations and universities. Likewise, during election seasons (are there any others?), polls commissioned by the Washington Post and ABC News move to the top of the news agenda, along with those performed by Rasmussen and Gallup. So, there’s a strong argument for ushering polls into our work. However, the origin of polling data

is a human interaction, which seems to make the practices of polling and sensing distinct. That said, we do acknowledge human agency in designing sensors, sensor data collection methods, and the analytic processes working with sensor data. Likewise, humans as sensors have been excluded from this report. Although we accept absolutely that humans make observations about the world through any combination of their five senses and can record the information, the lack of a mechanical process that can be interrogated and reproduced would seem to separate human sensing from technical sensing. Once again, we acknowledge counterarguments; advances in social science experimental techniques have made human observations more reproducible, while technical and mechanical sensors inherit design decisions influenced by human subjectivity. Nonetheless, human observations seem to have a different degree of controllability and specificity and are not as influenced by the macro-trends outlined above. For those reasons, we’re not researching human sensing. Perhaps the hardest exclusion has been software sensing. When marketing firms examine bit-torrent networks to research the popularity of movies and albums, their practice shares many characteristics with physical sensing. It is another intersection of new technologies with the demand to make observations about the world. It can produce massive amounts of interrogatable data. It can produce real-time information or information to be stored, processed, and disseminated. Still, there are differences as well; software sensing seems to be more concerned with the virtual world, whereas the practices we’re interested in here are more about observations of the physical world. But again that might be a false distinction: In our case studies we have examples of physical observations moving immediately into digital, networked information.

So, our borders of convenience that exclude these journalistic practices could easily be redrawn to welcome them in. Some of the characteristics of sensors we describe are shared with these practices we’ve left out. Some of the legal and ethical considerations of reporting with sensors apply, and the lessons and observations found in the case studies may be just as relevant. So, if you find any of our observations and lessons about sensing useful for other practices, take them with our blessing.

Sensors and Sensibilia: A Historical Survey

By Charles Berret The history of sensors is humbling in its scope. Humans have always experienced the world through sensors like our eyes and ears, but we also use an array of tools to extend those basic capacities, to monitor our surroundings, and to track phenomena that are otherwise imperceptible. These “tools” needn’t even be high tech. The proverbial canary in the coal mine is a sensor for poisonous gases; a blind man’s cane is a sensor for objects just ahead. Really, a sensor is anything that reacts predictably to the state of the world. Such a pat definition should raise a number of concerns about the construction of knowledge, and the authority embedded in what appears obvious, but this is a brief and broad survey of sensor technologies. It touches many cases, and regrettably skews toward Western ones, but hopefully this primer indicates the sheer scale of these instruments in the history of human sense-making. Archaeological evidence shows that humans built sensors even in prehistory. Scales have been unearthed in the ruins of the earliest civilizations, as it was essential to weigh goods for trade and taxation. Agricultural needs also led people to track the cycles of heavenly bodies with monumental markers. Neolithic circles like Stonehenge and temple complexes like Abu Simbel were massive instruments built to watch the skies for signs of spring thaw and autumn harvest, among other things. Similarly, the area near presentday Cairo, where the Nile splits into its vast delta, has traditionally been the

site where the river’s annual flooding was measured by a variety of instruments known as nilometers. A reliable warning for the rising waters could mean the difference between a year of abundance and one of hardship. It is worth noting that some of the earliest sensors were also aimed at supernatural forces. Oracles, charms, and portents were seemingly attuned to fates and spirits. Some holy figures specialized in reading animal bones and entrails to understand forces at work in the world, while throwing a supposed witch in a lake was once, it seems, considered a reliable sensor for the dark arts. Yet some efforts at divination actually prompted the development of what we would now consider scientific sensors. The first magnetic compass was mainly used to tell fortunes when it was invented in China during the Han Dynasty, but it would not be used as a navigational tool either there or in the West until about the 12th century C.E.1 Until then, travelers navigated by the stars, and of course the empirical study of astronomy was once highly entangled with the prophetic efforts of astrology. Indeed, it was the celestial circle of 12 zodiac signs that originated the geometric measure of 360 degrees, with each constellation assigned 30. Many cultures tracked the movement of constellations not only to chart the year, but also to find their bearings by night. The Greek astronomer Hipparchus (190–120 B.C.E.) is credited with inventing both the astrolabe and the armillary sphere, instruments used to predict the movement of heavenly bodies, to navigate, and to tell time. Later, the sextant and alidade were added to astronomers’ toolkits for measuring and charting the sky. Several Greek astronomers tackled seemingly impenetrable problems even with these limited instruments. One of the cleverest of these experiments was organized by the Greek polymath Eratosthenes (276–195 B.C.E.), who

estimated the circumference of the Earth through a single well-timed measurement. As the story goes, Eratosthenes learned that the sun would shine directly down a well in Aswan, Egypt, at noon on the Summer Solstice, meaning it was directly overhead—that is, roughly on the Tropic of Cancer, the closest point to the sun at that moment. So Eratosthenes measured the shadow of an obelisk in Alexandria at the same moment. Eratosthenes took the distance between Aswan and Alexandria, deduced the arc of the planet’s curvature between those two points, and thereby calculated the planet’s full circumference with remarkable accuracy. Although Eratosthenes probably estimated that distance by having a slave count his steps through the whole journey, the ancients also developed several instruments to measure distance and speed more precisely. The architectural theorist Vitruvius (80 B.C.E.–15 C.E.) described the schematic for an odometer, which would count a mile each time a vehicle’s wheels clicked through a certain number of turns. Nautical speed, on the other hand, was measured with a knotted length of rope attached to a plank of wood tossed overboard. As the ship moved, a sailor would count the number of knots to pass through his hands, and thus gauge the distance the ship had covered in a given span of time—often measured by an hourglass. This information helped the crew estimate its position in the voyage, however roughly, and steer the ship toward its port. After the fall of the Roman Empire, the center of science and technology shifted to the Islamic world. Scholars at centers of learning like Baghdad and Damascus made many advances in astronomy, in particular, in order to schedule prayer times and to plot orientation toward Mecca. Islamic scientists were also accomplished chemists, and meticulously documented the properties and transformational potential of different substances in search of the alchemical shortcut to gold. During the Renaissance, many new instruments and measures surfaced as Europe slowly emerged from the Dark Ages and saw, in certain pockets, the developing culture of the scientific laboratory. Among Leonardo da Vinci’s

(1452–1519) hundreds of inventions, he designed a hygrometer to measure humidity and an anemometer to gauge wind speed. But perhaps the most noteworthy scientific advancements during the Renaissance resulted from precision optics for telescopes and microscopes. With these tools, scientists were able to observe phenomena beyond the normal limitations of vision. What was once invisible or imperceptible came into the realm of rational scrutiny through these new instruments and sensors. The first telescopes were invented in Holland for use on land and sea, but Galileo Galilei (1564– 1642) adapted the design to observe the moon, stars, and planets. Galileo is also credited with the first thermometer, which he designed after noticing the regular expansion and contraction of some liquids in response to the ambient temperature. But the invention of the barometer by Galileo’s friend Evangelista Torricelli (1608–1647) is an especially interesting case. Although changes in air pressure are largely undetectable to us, they are a useful indicator of approaching changes in the weather. Thus, the barometer is perhaps the first instrument that did not simply augment or quantify a basic human sense like sight or touch, but rather produced an entirely new capacity through the use of a tool. The philosopher Blaise Pascal (1623– 1662) reputedly carried a barometer up the Puy-de-Dôme to demonstrate the drop in air pressure at higher altitudes. During the political turmoil of the Reformation, when travel was not only dangerous but expensive, many scholars corresponded and collaborated by mail. The astronomer Tycho Brahe (1546–1601) was an especially active organizer of networked data gathering. From his castle observatory in Denmark, Tycho printed and mailed observation forms to a network of astronomers spanning all of Europe. His compiled results were the most accurate and comprehensive star maps of his time. The French astronomer Nicolas- Claude Fabri de Peiresc (1580–1637) also made effective use of the postal system to coordinate observation of eclipses by a dispersed group of scientists. The collected observations allowed Peiresc to determine more accurate lines of longitude and thus plot more accurate maps.

Determining longitude presented a far greater problem at sea, so the Royal Society of London established the Longitude Prize with a sizeable reward of £20,000 for anyone who could solve it. The answer turned to developing a clock small and rugged enough to carry aboard a ship, but still accurate enough to keep time with a central clock at a known location. The clockmaker John Harrison (1693–1776) claimed the prize with his invention of the marine chronometer. By synchronizing this clock to the one housed at the Greenwich Observatory, a sailor could check the position of the sun against the known time in Greenwich, which originated the Prime Meridian standard in international timekeeping. Several other Enlightenment discoveries directly resulted from increasingly precise sensors. Joseph Priestley (1733–1804), the leading chemist of his time, believed that fire was caused by the release of a substance he called phlogiston—though no one had ever seen or even detected phlogiston. Antoine Lavoisier (1743–1794) finally discredited Priestley’s theory using a scale accurate enough to show that matter does not become lighter upon burning—as one would expect if it had really released its phlogiston—but instead becomes slightly heavier through oxidation. For this feat Lavoisier is considered the father of modern chemistry, though he still lost his head during the French Revolution. Another revolution, the industrial one, followed with a rush of inventions. Looking back on this period, the philosopher Alfred North Whitehead (1861–1947) once remarked that the greatest invention of the 19th century was really the method of invention itself. This was a muted critique of the relatively slow scientific progress amid the feverish technological push into modernity. The domestication of electricity in the 19th century was particularly transformative, and it marks a turning point in the history of sensors for several reasons. For one, electricity is the basis of the telegraph, the first instantaneous communication medium. Electricity is, of course, also a source of power, enabling sensors to be automated. Finally, many sensors today oper ate through transduction, the conversion of a physical quantity like sound or temperature to energy, often in the form of an electrical signal. Many of the sensors discussed below, and many that we still use today, are reliant on electricity in a variety of ways. The 19th century witnessed the arrival of technology that recorded images and sound. The first camera, which was unveiled to great fanfare in 1839, required long exposures for its chemical treatments to capture an image. But as inventors designed more sensitive film, the camera offered not only greater accuracy and detail than any drawing, but could also capture phenomena too fast and fleeting to be apprehended by the naked eye. Historians see this as the moment when mechanical sensors were first treated with greater credibility than the human observer, whose many biases and limitations could derail the objectivity of their findings. The photographer Eadward Muybridge (1830–1904), for instance, built an elaborate array of cameras to photograph a horse at regular intervals through the course of its stride. The photos were commissioned by the industrialist Leland Stanford (1824–1893) to settle a bet over whether or not horses fully leave the ground as they gallop. The resulting series of images captured each stage of motion, conclusively showing that the horse does indeed lift into the air as it runs. Here, the unique capabilities of photography settled an otherwise intractable debate. Likewise, the first sound recording technology enabled unforeseen possibilities to analyze, archive, and manipulate sound. Sound is so fleeting and inexpressible that we will never be certain what ancient languages and music were really like, thus the advent of recording it was a rather dramatic moment in the history of sensing. The first instrument that could record sound was Édouard-Léon Scott de Martinville’s (1817–1879) phonautograph, which produced etches to represent a sound visibly, but could not reproduce it audibly. These etches must have been novel and evocative, but they were clearly static and limited. Thomas Edison’s (1847–1931) phonograph, on the other hand, was the first to both create and play back brief

recordings from a wax cylinder. In both cases, the air pressure of the sound waves would directly move a needle to inscribe its mark. With the invention of magnetic tape in 1928, audio could be recorded in multiple takes, with sounds overlapping other sounds, to create pieces more complicated than the phonograph’s recording of a single moment. At the same time, medical instruments invented in the 19th century gave physicians the ability to monitor a patient’s pulse, respiratory rate, temperature, and blood pressure. The physician René Laennec (1781–1826) developed the stethoscope after watching children tap sounds to each other through a long block of wood. Ludwig Traube (1818–1876) realized that a patient’s fever corresponded to the trajectory of illness and recovery, so thermometer readings became a regular component of diagnosis and treatment. With Scipione Riva-Rocci’s (1863–1937) invention of the sphygmomanometer in 1896, blood pressure became the fourth vital sign monitored by physicians. Later, Willem Einthoven (1860–1927) was awarded the Nobel Prize for inventing the electrocardiogram to measure the heart’s electrical activity through a string galvometer. The first bedside monitor was used by the surgeons Aaron Himmelstein and Martin Scheiner in 1950 to simultaneously monitor a patient’s heart rate and electrocardiogram during an operation. Vital signs were plotted as waveforms on an oscilloscope, and alarms would sound if either one reached a dangerous level. These monitors were common by the 1960s, and soon their range of sensors expanded to blood pressure, respiratory rate, and body temperature, among others measurements. In the first half of the 20th century, astronomers too were probing for signals that we cannot detect naturally. Telescopes sensitive to radio waves, microwaves, or x-rays could scan and map energy from the distant reaches of the universe. NASA’s Search for Extraterrestrial Intelligence (SETI) famously distributed the vast scans of its radio telescopes to volunteers whose home computers would crunch data when they were not in use.

Astronomers have also used spectrometers to analyze the light emitted by celestial bodies. Subtle shifts in the color of stars, for instance, could reveal a great deal about their composition and activity. Edwin Hubble (1889–1953) reasoned that the red shift of some stars indicates that they are moving away from earth due to the continual expansion of the universe since the Big Bang. Likewise, the gravitational red shift of Mercury when we observe it from the opposite side of the Sun provided some of the first empirical evidence for Einstein’s theory of general relativity. Meteorology also benefited dramatically from the technology that emerged in the 19th century. In 1843, when many cities kept local weather data, Elias Loomis compiled that information to draft the first synoptic weather map depicting pressure fronts, wind movements, and weather conditions for the entire eastern United States on a single day. But this had been compiled from past data. He could only gather the data by post. But when the telegraph network began to link American cities two years later, current weather data could be gathered from a widely dispersed network of weather readings, and the first broad picture of weather systems could be stitched together from regular, recent data. In 1849, the Smithsonian Institution began gathering weather reports from a dispersed network of 140 volunteers, and by 1856 it had compiled and displayed a daily weather map of the country. As weather networks grew, meteorologists set up small, remote boxes called weather stations to shelter sensors like thermometers and barometers as they collected readings. For aerial readings, multi-purpose sensors called meteorographs were mounted to kites or hot air balloons. In the 1920s, the U.S. Weather Bureau dispatched a fleet of airplanes to gather weather data across the country. In 1928, the first radiosonde, an unmanned weather balloon, gathered high-altitude weather data and transmitted it back home via radio. And in the late 1930s, meteorologists began using doppler radar to map precipitation over entire regions.

In fact, the invention of radar and radio broadcasting are closely tied to weather experiments. Following Heinrich Herz’s (1857–1894) pioneering work on radio waves, the physicist Alexander Popov (1859–1906) inadvertently invented radar while he was trying to build a lightning sensor using radio waves. Popov noticed that each of the ships he used to gather readings were blocking each other’s measurements, but that this offered an oddly effective way to locate the other barges. Building on these findings, Gugliemo Marconi (1874–1937) invented the first radio communication system as a means to send telegraph messages wirelessly. The same pings that we associate with a radar screen would, in this case, beat to the rhythm of Morse code and send the message out over the air. In this way, Marconi’s wireless telegraph was strangely kindred to the wi-fi and cell phone transmissions we still receive on the radio spectrum. The staccato volleys of telegraph tones were quite literally digital, and they share many qualities with the languages and encodings that circulate through today’s electronics. Given the many uses we still have for the radio spectrum, it is worth recalling that old technologies rarely go away. Weather vanes still perch on roofs to tell the direction of the wind, mercury thermometers can detect a fever in a pinch, and the magnetic compass is still an effective navigational tool. Digital instruments are more common today, and in many ways more useful for data analysis, but the story of sensors is vastly historical. Although this section stops well short of the present day, it should illustrate that sensors have played a massive role in human history. Much of what we know about the world, we know through sensors that extend and quantify our natural capacities. We can say with certainty if it was warmer yesterday than it is today. We can reckon when it is midnight in Delhi. We have heard the voice of Winston Churchill. Sensors are also at the heart of communications media that enable us to gather and distribute information. Many of the researchers discussed above were only able to make progress in their work once they could gather data

from a dispersed group of collaborators. Sensors enable us to investigate what we simply cannot see, hear, or touch. These instruments have quite literally provided us with new senses, and they are, as a result, the most difficult to scrutinize.

The Characteristics of Sensors

Journalists considering whether to include sensor data in their own reporting may want to evaluate their story, their goals, and the potential data they need. This section is intended as an aid for readers to understand the differences between sensors and the range of characteristics they can have—thereby helping them match the best tools to their needs. It should also be useful for reporters who are looking for data; this set of continuums may help them broaden the range of places they go looking for sources. The final use might be for readers who want to examine other people’s work with sensors, to help them analyze whether the purported conclusions can be supported by the underlying data production process. The characteristics labeled below are only the ones that seem most important and commonly applicable. It is not useful for us to work through every potential characteristic of a sensor system. By way of example, most journalists will want to consider their sensors’ degree of accuracy and precision, a few will need to consider power use, and almost none will need to consider how old their sensors are. These characteristics may be a function of an individual sensor, or a whole sensor system.

Measurement Qualities These characteristics primarily concern the data that a sensor produces.1 Sensitivity to Target Phenomenon Simply, the relationship between the amount of phenomenon the sensor is intended to detect and the amount of the sensor’s output. Sensitivity to Interference The degree to which a sensor’s detection of the target phenomenon is influenced by other factors. In most cases, users will want their sensor systems to be insensitive to interference. Precision The degree to which a sensor can produce a data that is exact. Range The degree to which a sensor can detect very little of the phenomenon, up to a lot. For example, some accelerometers may have a range of only -2 times gravity, to +2 times gravity, whereas others have a range of -4/+4 or greater. Linearity The degree to which a sensor’s output is consistent across the whole of its range. A temperature sensor has a high degree of linearity if it records temperature to within 1 degree at -30 and +30 (and everywhere in between).

Resolution This quality has two important facets: spatial and temporal. A sensor system with high temporal resolution will produce data with lots of values in a given time period. A sensor system with high spatial resolution will produce data with lots of values for a given area. Operational Qualities These characteristics may act upon the previous set of qualities, but may also affect how practical it is for journalists to use the sensors, or access the sensors’ data. We include these qualities here because of their relevance to our case studies. Maturity A sensor system may be mature if it has been widely used, thoroughly tested, and is not undergoing rapid functionality development. Sensor systems that are immature are less likely to be suitable for applications where users need reproducibility and reliability. A recently designed prototype water quality sensor, for example, is unlikely to produce data that can withstand challenges from stakeholders, or be used in courts to prove water is unsafe. Ownership Sensor systems may be owned and/or controlled by individuals, governments, or private entities. Ownership may be relevant because it affects whether journalists can access the data, where sensors may be placed or moved, and which sets of laws govern the information the sensor is permitted to collect. It will be difficult for journalists to access data from sensors owned and operated by private companies. The operators of governmentowned sensors may, for example, need to consider the United States Constitution’s Fourth Amendment restrictions on unreasonable searches.

Autonomy Sensor systems can require various levels of proximate, immediate control to operate. A handheld x-ray soil contaminant sensor is under close control, whereas a camera mounted on a drone flying between preset waypoints is under less immediate control. Operating Distance Various sensors are designed to work at different distances from their subject. A Fitbit activity monitor only works if it is directly touching the subject, whereas a satellite collects information at a vast distance (especially if it is turned away from the Earth).

What Can be Sensed

Attendees at sensor reporting workshops and panels often ask, “What can be sensed?” Unfortunately, that is a simple question with a complex answer. There are a number of sources available to introduce journalists to sensing possibilities. For electronic prototyping, lists of sensors can be found on parts retailers including DIY stores Adafruit and Sparkfun, or online stores like Mouser. Wikipedia also has a “list of sensors” page. The electronics retailers divide their catalogue, into categories to guide buyers. Examples include motion, sound, scanners, touch, and biometric. Wikipedia’s categories, on the other hand, mix technologies with applications; one “type” is chemical, another is automotive. In any case, browsing those sources can help readers start to see what physical sensor parts they can buy is theoretically possible. However, note two points when it comes to the question of “what can be sensed.” First, lists of sensors are long and defy consistent organization. Second, logic and imagination have as much to do with answering that question as do the technologies. In the case study to come about the Sun Sentinel’s Pulitzer Prize-winning investigation, sensors on tollgates registered times and radio frequencies emitted by tags on cars driven by police passing through known locations—from which the journalists derived identities, their speeds and concluded ’criminality.’ This is all to say that while no one suggests that the journalists used a criminality type of sensor, or that Wikipedia should include a ’criminality’ section on its sensor page, but sensors still helped prove that Florida cops were breaking the law.

Likewise, a Washington Post story based on ShotSpotter data—gunshots sensed via sound—relied on the fact that explosions in a gun barrel cause air-pressure changes (also known as sound waves); these were converted into digital signals by arrays of microphones and pattern-matched by computers to produce records of the gunshot locations throughout Washington, D.C. The point here is that rather common sensors can feed data into processes that apply various computations of complicated physics and produce higher-level applications. Journalists have conducted further logical analysis and combined other reporting processes to derive some insight into the world. At each of the steps—between physics, application, and insight—engineers and journalists make decisions that affect what can be measured, derived and the analysis that can be made. So, the question of ’what can be sensed’ has different answers depending on which step in the process you are discussing. The answer can also change as journalists apply effort and immagination.


The Second Section: Case Studies


Case Studies: An Introduction This section will give readers a grounding in the current practice of sensor journalism. Some of the following are cases of journalists using sensors, activists using sensors, journalists using things that seem a bit like sensors, and professionals piloting flying robots with camera payloads. Each case study included here has practices to learn from. We see examples of the techniques and equipment that journalists have discovered to report their stories. We see the processes they developed to protect the communities in which they work. We see journalists navigate the tricky questions of what it takes to produce accurate data, and whether that’s actually what they’re trying to do. (That’s not as straightforward a question as it might appear.) On an operational level, we see some indications of the budgets involved —always of concern for newsroom managers but increasingly of interest to frontline journalists as well. The incorporation of sensors into journalism (and its adjacent fields) has taken a few distinct styles. The first is to design one’s own sensing process to produce data from mature, commercially available equipment. Another is accessing data from existing sensor sources. A third is designing prototype sensing systems to produce data. Dina Cappiello, working at The Houston Chronicle, and Alison Young of USA Today, had specific topics they wanted to investigate. As part of their reporting process they went looking for sensors they could personally operate to produce data to power their stories. At the Sun Sentinel, Sally Kestin and John Maines negotiated for data from tollgate sensors when they found it was the only way they could prove a com monly held belief that Florida police forces were rife with speeding cops. Journalists at the Washington Post also negotiated for sensor-derived public records. They’d found out about a network of audio sensors operated by the Metropolitan Police Department in Washington, D.C., and wrote their story partially as an analysis of what the data described and partially as an explanation of the police’s opaque crime-fighting tool. Two of our case studies cover projects where media makers have also become hardware designers. One, WNYC radio’s archetypal sensor project, The Cicada Tracker, was started by a data journalist named John Keefe before it was adopted by a community of electronics makers. The other, Public Lab’s activist environmental hardware development, might not even be journalism—but these lines are blurring and it is a fascinating movement so we have levered it in. We’ve included a case study about the NPR program Planet Money’s use of a drone, and its unintended camera sacrifice—both byproducts of technology’s improving bang-to-buck ratio and sensor miniaturization. The format of these case studies will, we hope, satisfy readers who are familiar with the projects and those who are reading about them for the first time. The cases start by describing what happened: what the story was, how it was reported, the kit, the facts, and how it was published. Then, you’ll find a section outlining the distinctive and notable elements of each case, before reading some takeaway lessons for the journalism industry. This collection of case studies is by no means exhaustive. Although we have included drones, that is the highest altitude we venture. Out in space orbits a whole fleet of satellites, public and private, carrying a bevy of remote sensors. Newsrooms currently use satellite-derived data in their maps, and at least one newsroom has current investigations that leverage infrared sensor data and time-series of visible light images.1 In the early phases of the research underpinning this report, we expected to have more examples of

custom-built sensor projects to study. However, through our own activity and through observing the mainstream of the profession, we found fewer than expected examples of journalists building their own sensors. Costs and the difficulty of producing accurate data have sunk journalistic projects of that type. While researching and writing these cases, three key themes emerged. Journalistic sensing is often intertwined with community. The physicality of sensing tends to mean that reporters have to work actually in their communities and must consider how their activity will interact with the people living where they are taking measurements. Second, the journalists in these case studies learned as they went. They found out about technology and researched its processes. Even the reporters who had formal training or long experience in their specialized beats had to study up to get the story right—and not just on the subject of their articles (which journalists almost always do) but on techniques and practices from professions outside their own. And lastly, but crucially, these journalists were not collecting their sensor data in isolation. Not only did they add context to the data, to make their audiences care, they rendered colorful pictures of the affected people.

Houston Chronicle—In Harm’s Way

In the words of environmental journalist Dina Cappiello, “Houston bears the environmental costs of the country’s appetite for fuel.” In 2002, while Cappiello was researching story leads for her impending move to the Houston Chronicle, the south Texas economy was dependent on the petrochemical industry. Unlike most large cities, Houston had no zoning laws in place to separate its residents from oil refineries and Texas’ air pollution limits were orders of magnitude looser than other states’. Poor neighborhoods sat right next to factories that periodically released toxins in plumes of black smoke. They were widespread enough to have attracted their own label: “fence-line” neighborhoods. Cappiello was hearing stories from Houston residents about chemical rainfall, followed by visits from oil-company employees who would offer free car washes. One time after black soot fell, according to her sources’ stories, the companies had rushed around buying up children’s toys that had been left out in backyards. But her reporting found that all evidence of abnormal air pollution was anecdotal. Despite fence-line residents saying they got nosebleeds and smelled a stench in the air, the companies maintained that regular monitoring showed pollution releases were legal and posed no threat to health. The Texas Commission on Environmental Quality (TCEQ) agreed.

In the Houston Chronicle’s yearlong investigation that followed, Dina Cappiello developed an innovative data production process that galvanized attention, sparked political action and industrial changes, and provided a template for future environmental pollution reporting. The missing part of the story, as she saw it, lay in the gray area between the residents’ anecdotes and the denials from industry and state regulators that there was anything to worry about. Cappiello holds a bachelor’s degree in biology from Georgetown and a master’s in earth and environmental science from Columbia University. That education gave her a respect for hard data. So when she wanted quantified and specific information about what was in the air around those fence-line communities, she defined a core question: “What was in the air, how much, and was it enough to put people at risk?” The EPA maintains a database of toxic releases. Factory owners report their emissions into the Toxics Release Inventory (TRI). It includes records of the planned emissions that are the day-to-day byproducts of normal operations, combined with data about their unplanned chemical releases that often follow some accident or mistake inside industrial plants. Cappiello could use the TRI data to calculate the total amounts of air pollution released in the Houston area, but she also wanted the concentrations—a key measure scientists need to research health risks. The official monitoring took samples only once in each six-day cycle and TCEQ installed its monitoring stations away from the fence lines (although the permanent stations were sometimes supplemented with mobile monitoring trailers). Cappiello wanted data that had a higher spatial resolution; so she could find out what was in the air right where individual residents lived. Her reporting focused on four neighborhoods—one each in Houston (Manchester/ Allendale), Baytown, Freeport, and Port Neches, where the nearby industrial plants posed the highest risks of harmful exposure. Throughout the Gulf Coast, community activist organizations had already developed

DIY data collection techniques, but Cappiello couldn’t trust they would stand up to the government and industry scrutiny she expected once her story was published. To develop the Chronicle’s process, Cappiello consulted Dr. Thomas Stock, an associate professor of public health at the University of Texas Health Science Center at Houston. His input could only be informal unless the team was willing to go through the university’s ethical review process for experiments with human subjects (traditionally a long and arduous one). The approach Cappiello settled on borrowed techniques from the oil industry’s safety practices. Workers in the industrial plants around Houston wear chemical monitoring badges called Organic Vapor Monitors, made by the 3M Company. They’re about the size of an ID card but round and designed to hang off a shirt pocket or jacket lapel. They come in a small aluminum can and when a pull-tab lid is peeled away, a charcoal pad at the back of the badge starts absorbing chemicals from the air, a process called passive diffusion. Depending on what one is trying to measure, a badge can collect chemicals for days at a time. At the end of the test period, the badges are capped and sent to a lab, where technicians extract the contents and run it through a gas chromatograph/mass spectrometer. Bought in bulk, the badges themselves cost about $20 each, but analysis is extra: the Houston Chronicle paid between $100 and $120 per badge to collect data on 31 chemicals, of which 18 were potentially toxic. Cappiello says the project’s cash costs hit $20,000, after covering the badges, the processing, and the reporters’ travel and accommodation. The story Cappiello hoped to write stood a high chance of attracting the ire of the local industry, so by deploying the same equipment that the oil refineries used to protect their workers, she hoped to be able to deflect at least one line of attack. In each of the four neighborhoods she would need to find 25 places to set up the badges. The most powerful angle of her story was the toxicity of the air around the residents’ houses, which meant she needed to recruit volun teers who were willing to hang the badges from trees, play equipment, or awnings outside their homes. The newspaper wrote a letter in English and Spanish, which Cappiello and her intern took with them as they went door to door, sometimes delivering their pitch in person, sometimes leaving the letter behind if no one was home. A locally active politician from Houston City, Councilwoman Carol Alvarado, helped recruit the volunteers in Manchester. Cappiello says she asked a lot of the residents—they had to agree that their names and photos could be published. In some cases a breadwinner worked for the factories, in other cases the residents were worried that their home’s value would drop if it turned out that the surrounding air was poisonous. Across the four neighborhoods where Cappiello reported, she recruited 84 residents to hang badges outside their homes. The Chronicle gave detailed instructions that badges needed to hang at head height, where adults breathed most of their air, but away from grills and carports that could interfere with the results. Another badge hung on church grounds. Cappiello hung 15 more in public parks and playgrounds, taking care to keep them out of the way of children. She says because the land was public space they didn’t need to ask permission from the local councils (although an unknown informer tipped off the FBI, which meant she spent a phone call convincing the FBI her work was benign). Based on the prevailing winds in each neighborhood Cappiello planned to have 20 badges downwind of the plants and five upwind—although the winds turned out to be significant in only one area when the tests were run. She also hung blanks in neighborhoods away from the fence lines, to establish a baseline. In each community Cappiello instructed the residents to open the cans containing the badges at the same time. Within a two-hour window at the start of the test period, the Chronicle visited each house to check the monitors were set up correctly. At the end of each community’s 72-hour test period, the reporters collected the badges, put them into coolers to reduce any subsequent reactions, and sent them to Dr. Thomas Stock’s laboratory at the University of Texas School of Public Health for processing.

The individual badges’ datasets were handed back to the Chronicle for analysis and interpretation. For each chemical Cappiello calculated maximum, minimum, average, and median concentrations, having discarded data from badges where water droplets indicated potential interference. When she had done her calculations, she sent her data and methodology to a panel of six environmental scientists from government departments, private companies, and top-tier universities. Each of the volunteers who’d hung monitors around their homes got a printed report detailing what the badges detected and what it meant for their health risks. The Chronicle’s story, published in January of 2005, focused on three cancer-risking chemicals in particular: Benzene, 1,3-Butadiene, and Chloroform, all chemicals they found in higher concentrations in the tested areas as compared to in Texas’ typical urban areas. The Chronicle’s results fell within the ranges found by the official state monitors close to the four neighborhoods. That gave Cappiello confidence that her data was plausible, but also indicated that the air contaminants did not exceed the legal limits for Texas. At this point, the analysis shifted to deal with the complexities of known health risks in correlation to the state’s regulatory limits and the limits imposed in other parts of the United States. In all four of the neighborhoods the Chronicle monitored, the median concentrations of Benzene were above the EPA’s level where continual exposure increased the detectable risk of cancer.1 Almost all of the individual badge’s results were above that level as well. In Manchester, the maximum concentration was 27 times over that level.

For Chloroform, two neighborhoods’ median levels fell above the EPA risk level, and two fell below although most of their ranges were above.

Both Dina Cappiello and Dr. Thomas Stock agree that the methodology the Houston Chronicle used was robust in the context of journalism, but didn’t approach the standard found in a peer-reviewed health sciences journal. Cappiello acknowledges the critique that, for a high level of confidence, testing should be done at least four times over the course of a year. Likewise, Dr. Stock says the Chronicle’s testing showed what was in the air in the neighborhood, but didn’t incorporate indoor air quality—where residents breath some of the air that makes up their individual dosage of harmful chemicals. Dr. Stock says that the Chronicle’s project did, however, raise awareness of the problems in Houston’s air quality in a way that academic studies rarely do. In the weeks following publication, the mayor of Houston, Bill White, funded a task force of experts to investigate Texas’ air pollution and its health effects. Its work became the basis of new academic study. Mayor White also announced a plan to set up more air quality monitors in fenceline communities. The Texas Petrochemicals company signed an agreement with the city to reduce its butadiene emissions, and by December of 2005, the Texas Commission on Environmental Quality had increased its air pollution research and monitoring, especially along the fence lines. The reporting project also had an impact in the journalism industry. Cappiello’s process inspired Blake Morrison of USA Today when he was planning an investigation of the air quality around schools. Five years later, he used the same organic vapor monitors as did the Houston Chronicle, supplementing that data with active air-pump monitors and monitors which use ultra-violet light to detect air contaminants. In the 10 years since Cappiello’s investigation, air quality has been a hot political topic in Texas. The state’s environmental regulators have continued to set different standards from the federally operated EPA, and that agency continues to bring cases against companies that release hazardous chemicals.

Distinctive and Notable

  • This was non-electronic sensing. Dina Cappiello’s data production method shares characteristics with how contemporary sensing is understood. She deployed chemically reactant badges, tools that recorded aspects of their environment and could then yield information. The process extended human reporting abilities by making the observations specific and quantified, and allowed her to collect observations in many places simultaneously. Electronics didn’t come into the process until the badges were sent to Dr. Thomas Stock’s lab.
  • The Chronicle’s data was close to official data. The Houston Chronicle’s initial hypothesis was that the official government air pollution monitors were too spaced out, and in the wrong places, to record dangers to the fence-line residents. While that may have been true, the data Cappiello produced when she took readings with a higher spatial density fell within the ranges produced by the more spaced out, official monitoring stations. Furthermore, all the sensors, government and journalistic, recorded levels of air pollution that were within Texas’ legal limits, even if they were higher than other states’ limits. However, the fresh set of data—evocatively linked to individual people, whose faces and biographies the Houston Chronicle showed—allowed the newspaper to kindle a debate about whether the legal limits were safe and appropriate.
  • The community’s efforts were huge. Community management formed a significant part of Cappiello’s reporting process. She cultivated a contact who helped recruit volunteers in Houston. She produced dual-language information sheets and coached volunteers in the monitoring process. Throughout, she remained aware that she was asking volunteers to help her produce information that might have had significant consequences: lowering the value of their homes and introducing health fears. Community management appears to be a necessary skill common to other journalism that incorporates environmental sensors.
  • To avoid lengthy “human subjects” review, Dr. Stock’s advice had to be informal. When the Tow Center discussed the Houston Chronicle’s project with Dr. Stock, who advised Cappiello on how to use the monitoring badges, he took pains to point out that his advice was informal. If he had formally partnered with the Chronicle, he would have been obliged to run the project through his university’s ethical review process for investigations concerning human subjects. Typically, human subjects review processes take a number of months, can require training for every team member who comes into contact with a volunteer, includes a baseline assumption that all volunteers remain anonymous, and that only team members named in the plan have access to their data. The review processes are designed to ensure that research does not harm participants. However, they are not set up to integrate with the requirements of journalism and publishing.
  • They hung monitors in public parks. The Houston Chronicle journalists made an assessment that they didn’t need to get permission to hang their monitor badges in public parks. However, setting up monitors in public, unprotected locations was not without its drawbacks: According to Cappiello, they did lose “a couple” of badges over the course of the project. The reporters, one of whom was wearing a Muslim headscarf, also attracted the attention of the FBI, which called Cappiello in response to reports of suspicious activity.

Lessons for the Industry

  • Analyze other data sets to narrow down your investigation. Before deciding where to place her monitors, Cappiello inspected the EPA’s toxic releases inventory database. That helped her focus her reporting on four particular neighborhoods.
  • Assure volunteers access to the data. Cappiello ensured that the residents who hung monitoring badges around their houses had access to the lab results once the badges had been processed. Her decision to do so appears to anticipate the ethical debates regarding data and access that have become more urgent in subsequent years.
  • Be advised that those affected by your data will want to know the bad news. Cappiello, when contemplating what she was asking of volunteers, realized that she could potentially produce data that threatened residents’ house prices or imply that they should move to protect their family’s health. If they worked in the refineries themselves, Cappiello’s story could cause problems for their employers. While some residents refused to participate, many of her volunteers adopted the attitude that if there was a problem, knowing about it would empower them.

Public Lab – Homebrew Sensing

The origin story of Public Lab, a non-profit community collective seeking to investigate environmental concerns with DIY tools and techniques, is deeply seeded in the BP oil spill off the coast of Louisiana in 2010. When an explosion on April 20 sunk the Deepwater Horizon oil rig, killing 11 workers and ripping open the undersea wellhead, the accident caused the biggest oil spill in history. As the crude oil polluted the seawater and coastline, journalists from across the United States and around the world flocked to report the massive environmental story, threads of which were still making front pages four years later. The story had traction for a number of reasons: Facts and analysis were highly contested, information was selectively released and journalists had to fight for access to the places where they wanted to report. Some of the people at the core of today’s Public Lab, Liz Barry, Jeff Warren, and Shannon Dosemagen, were among the volunteers who rallied around the Gulf Coast, helping with the cleanup and hoping to hold BP accountable. They came at the task from various backgrounds in environmental activism, innovative technology, and media and civic participation. Although none identified as a journalist, they were trading in information and their motivations approximated those of investigative reporters. In 2010 the group’s particular focus was working with local groups to record where the oil had spread; a crucial piece of the puzzle in understanding how much damage the spill would cause and how the cleanup money should be spent.

Since then, Public Lab has started to define itself, evolving into a community aiming “to change how people see the world in environmental, social, and political terms.” They encourage the use of inexpensive, accessible equipment to collect information as a first step toward civic participation. Around the Gulf Coast in 2010, this meant weather balloons, kites, and cheap digital cameras. Later, it meant working with open source electronic air and water quality sensors. Community and Journalism Innovation Public Lab’s position in the news media landscape, however, is hard to define. Although the word journalism doesn’t appear in its mission statement and its mailing lists attract more activists, engineers, scientists, and teachers than they do reporters, as of April 2014, Public Lab has won $850,000 from the journalism and media innovation section of the Knight Foundation.1 Those Knight News challenge awards account for about 55 percent of the group’s cash funding. Arguably, Public Lab fits within the new framework of post-industrial journalism, where news-like information flows through many publics and can take any number of routes to an audience. Some information reaches audiences through mainstream media organizations, but plenty more circulates on email lists, discussion boards and social media. A co-founder of Public Lab, Shannon Dosemagen, told the Tow Center that one of her current foci is to improve how Public Lab data can be used by civic institutions, such as the courts and the legislative process. For the purposes of this case study, to fully understand how Public Lab’s environmental sensing relates to journalism, it is necessary to understand the Public Lab community itself. Jeffrey Warren, a co-founder and the group’s research director, says that when he introduces Public Lab to someone he is meeting for the first time, his description shifts around: He may say that Public Lab is a “platform” or “a network of communities.”

A common thread in explanations of the organization is that Public Lab avoids a top-down structure, or a singular drive toward a known outcome. Public Lab’s operations facilitate people meeting their own needs; often, but not always, those needs concern environmental activism. However, the organizers believe that if people are working in a connected way they will have opportunities to learn from one another, share tools, and realize unanticipated benefits. Aerial Mapping Versus Official Data In 2010, when oil was spilling out of Deepwater Horizon, those gnarly, high concept questions were yet to be raised, let alone answered. In fact, Public Lab didn’t even exist. The activists’ immediate concerns were twofold: In the short-term they wanted to know where the oil was reaching the land, so they could focus cleanup efforts. Later, that information might also feed into an assessment of liability and damages. At the time when the spill was at its worst, Louisiana authorities were blocking photographers from accessing many parts of the shoreline. Airspace restrictions also banned flyovers below 3,000 feet, an altitude too high for news helicopters or planes to collect imagery with enough resolution to show oil hitting the land. Public Lab’s methodology for documenting where the oil made landfall combined elegance and scrappiness. They relied on an old, simple technology— weather balloons and kites—hanging digital cameras to their craft, stabilized with fins made out of plastic soda bottles. They floated the contraptions above the shoreline, holding onto them with thin kite strings. As the balloons flew, the digital cameras snapped photos under the control of automatic timers, capturing images at pixel-per 3 cm resolution. After, they pulled the balloons back in to retrieve the memory cards, loaded the files into a specialized photo joining tool called Mapknitter, and stitched them

together to show stretches of coastline. The aerial photos showed black oil stains at many points from just south of Lafayette, La., through to Perdido Point, Ala. One of the rationales for journalists producing their own data (using sensors or any other method) is that official sources cannot always be trusted; perhaps because they are incomplete, or they are wrong. Mat Lippincott, who designs hardware for Public Lab, points to a particular case in Louisiana where he says the official record of where the oil made landfall is wrong. After the BP oil spill, the Emergency Response Management Agency (ERMA) built a geographic database, rating oil pollution against a six-point scale for dozens of coastal stretches from Louisiana to Florida. That database became a source for maps published by The New York Times, the Guardian, and plenty of other outlets. The official ERMA record, based on its contractors’ observations for July 22, 2010, says that in Wilkinson Bay, 20 miles south of New Orleans, floating boom-barriers kept the oil away from the shore.2 By chance, the Louisiana Bucket Brigade, one of the local environmental groups to which Dosemagan belongs, was also operating balloons and kites in that location, on that day. Their photos show heavy oil covering the reeds and marshes and broken booms that were intended to block the oil from reaching land.3 (Although ERMA’s underlying data might be contested, no news media’s maps we saw showed data at a high enough resolution to see whether the inconsistency Lippincott cites was relevant or included in their publishing.) Balloon mapping has since become a common part of Public Lab’s data collection. They have used it to document sites of environmental conflict or various “occupy” crowds.

The Next Step: Tracing Source Later in 2010, the questions for Gulf Coast activists shifted from mapping the spread of oil to establishing its source. Although at the time BP and local governments were saying the cleanup was mostly complete, beach walkers were still seeing tar balls and oil residue. BP’s response was that the oil could have been from any number of minor spills that happen throughout the Gulf Coast. According to Jeffrey Warren, the Public Lab people had no knowledge of how to identify and fingerprint hydrocarbons. They started their research in the typical way, by typing this question into Google: “how to identify oil,” and the results that came back pointed them toward spectrometry—a basic technique that has remained in use since its origins in the early 18th century. At its most basic level, spectrometry relies on the fact that different substances absorb different amounts of colored light. After shining light through a sample, the amounts that remain produce a signature that can be compared to a catalogue of known samples. It is essentially a fingerprinting process. Although various spectrometers have specific components that allow the user to see the signature more clearly and do the comparisons more aptly, the basic concepts remain consistent. In the case of oil from the BP spill, the activists’ theory was that they would be able to take a sample of the black sludge on a beach, compare that sample’s signature to that of the known oil spilling from the Deepwater Horizon accident, and show liability. Perhaps unsurprisingly, that level of accuracy is hard to achieve, and in the case of oil from the Deepwater Horizon Spill, did not get close oil spill. Nonetheless, its idea of designing a low-cost spectrometer survived. Over the next three and a half years, members of Public Lab have been producing new versions of the hardware, all within the bounds of their underlying DIY philosophy and design parameters. Today, they have two main designs:

a small plastic version that clips on to the back of a smartphone and uses its camera to record the spectra, and a larger “desktop” version, which incorporates more people’s design and development efforts. They aim for simple construction and incorporate materials that are ubiquitously available. For example, they use a section of DVD as their diffraction lens and their desktop version uses a standard household plumbing conduit as its body. The Open Source Hardware Approach Although the original demand for the spectrometer was the Gulf Coast spill, Public Lab’s philosophy is to support whatever application their community finds for their equipment. Josh McIlvain, who completed a physics undergraduate degree from Kansas State University, used the spectrometer to analyze the components of laundry detergent, detecting dyes in a supposedly “free and clear” brand. Environmental activists who suspected their local Louisiana industrial plants were releasing harmful contaminants have developed a plan to use the spectrometer to analyze flares from an oil refinery and a lubricant factory. But mostly, the uses shown on Public Lab’s community upload section tend toward experimentation and exploration. Public Lab’s approach has been successful in attracting funding: It ran a Kickstarter campaign in 2012 to improve its DIY Spectrometry Kit and raised $110,538, 11 times its $10,000 goal. In the Knight News Health Challenge of 2013, Public Lab won $350,000 for its Homebrew Sensing Project. The pitch to the Knight Foundation paints a picture of a decentralized network of contributors using citizen science approaches to document and address “environmental health threats and their impacts.” Public Lab’s open operating model extends beyond facilitating unanticipated uses of its hardware. It also supports community members’ development of their own equipment. One such example is a prototype called “The Riffle,” an electronic water sensor designed to be deposited in a stream to collect data on water quality and pollution. The device designers, Ben

Gamari and Don Blair, are both fifth-year biophysics Ph.D. candidates at the University of Massachusetts, although this work is not part of their formal study. The current prototype of the Riffle logs temperature, water pressure, and conductivity, which can be broad indicators of abnormal activity; especially high levels of agricultural fertilizer runoff or releases from industrial plants. While those measurements are basic, Gamari and Blair say that they want to make it viable for citizens to continuously monitor water quality at very low cost. The Riffle is designed to use mostly household plumbing parts and cheaply available Arduino-based electronics, although one specialist component needs 3D printing, a production method that they hope will become increasingly mainstream. Since winning the Knight News Challenge in 2013, Public Lab has picked up more philanthropic funding, some for basic operations and some for specific issue-based campaigns. The Schmidt Family Foundation has funded Public Lab to work with Wisconsin residents and monitor air quality around frac-sand mining, a production process that can send fine-grained particles into the air and put citizens at risk for respiratory problems and possibly lung cancer.

The Challenge for Citizen Science, and the Roadmap However, Public Lab’s work, especially its spectrometry and electronic sensing, does attract skepticism. In comparison to lab-grade equipment, Public Lab’s current hardware is extremely immature, and while it does produce data, it is not clear whether that data can be compared to known health or regulatory standards to the degree that would stand up in court or withstand a contest from polluters. Likewise, the spectrometer’s sensitivity, precision, and range are yet to meet the requirements of measuring samples with the kinds of pollution concentrations found in many real world environmental situations. Dr. John Feighery, an environmental engineer who consults for the World Health Organizataion and UNICEF on water quality and managed the development of water monitoring hardware for NASA’s International Space Station, describes the difficulties in Public Lab’s spectrometry use. “The challenge in water analysis is both detecting very low levels and with a large number of other compounds present. These are the typical conditions in environmental samples. If you think of the spectra of one compound as like a fingerprint, then imagine dozens of very faint fingerprints all on top of each other on the same spot, you start to get the picture.” Until Public Lab can produce data showing detection of a contaminant of interest in a realistic environmental sample, Dr. Feighery says he will remain skeptical. He told the Tow Center, however, that he admires Public Lab’s work, but thinks its current spectrometer development seems more like a hobby than a practical device for fieldwork. Jeffrey Warren is aware of experts’ reservations, and his response to it hinges on the idea that Public Lab’s activity does not have the kind of singular goal to which scientists or journalists are accustomed. “We talked a lot about this early on in the founding of Public Lab. A lot of people asked, ’Are you focused on the benefits of people being involved in the process, where the involvement itself is a transforma tive approach to creating environmental science, or are you focused on the outputs: the things you can prove, the legal decisions you can influence, the regulations that you can enforce? Or, are you interested in the media attention that you can generate?’ And I know it’s not a satisfying answer, but the answer is, Áll of the above.’” “For us to have a citizenry that is capable of affecting change through the collection of data, through the advocacy based on that data, through the strong, productive relationships with government and even industry, they’re going to need to be involved in the process in more than a token way. They’re going to need to learn about the techniques.” Other people within the Public Lab community say that their hardware development process, if given time, should show good results, and cite the open source software movement as an example of how tools’ quality and utility can be built over the long term. The Riffle designers, Gamari and Blair, who use the open source programing language Python in their labs, say the goals for their water quality sensors have always drawn on that experience. Blair says, “Part of the vision is having something like [open source software community] Stack Overflow for all of the instrumentation that people are using in science.” Distinctive and Notable

  • There’s activism, then there’s journalism. The Tow Center has included Public Lab in this report largely as a result of the funding it has won from the Knight News Challenges. While its founders don’t self-identify as reporters and their activity looks nothing like traditional journalism; they don’t write mainstream news stories, make video programs or broadcast radio shows, many people in the Public Lab community appear to share motivations with journalists—to uncover information and put it to use in the civic space. Mainstream media lost its monopoly over the means of media production, publication, and distribution years ago. One result is that the distinctions have blurred between journalism and everything else. Public Lab’s activities are a case in point.
  • Data utility, and how data accuracy informs it. Public Lab projects produce data with highly variable utility. Their aerial mapping techniques can produce photographic evidence that is reliable and believable. For journalists, these techniques may be useful as a source of content and data that is alternative or additional to official sources, especially while drones are unsafe, illegal, or impractical. Their other hardware projects, the spectrometers, the Riffle, and the near infrared cameras, are yet to produce data meeting a standard of accuracy where it can be compared to legal limits or health standards. For journalists who are working on investigative projects or stories where an issue is environmental safety and legality, the immaturity of Public Lab’s tools should give pause. However, Public Lab members say that the data they produce still has value as an early warning that better monitoring is needed, or to make illustrations that attract media and public attention. They also say that the very act of participating in data collection educates and engages a citizenry to become environmentally concerned and politically active.
  • Newsrooms could frame Public Lab content as the UGC of data. Progressive newsrooms have become increasingly sophisticated in how they incorporate user-generated photos and videos into their journalism. In broad terms, attitudes evolved from a lack of interest, to suspicion, through enthusiasm to judiciousness. Big news companies such as the BBC have verification regimes and the startup company Storyful built a $25 million business on providing that service to many newsrooms. So far, we’re not aware of any newsrooms incorporating community-produced environmental sensor data into their UGC harvesting. This may be an opportunity.
  • Public Lab has engaged the maker movement. On its mailing lists and wikis, Public Lab hosts skilled electronics enthusiasts who are keen to use their abilities to contribute to a common good. They design hardware, test prototypes, give feedback, and suggest developments. This appears to be a positive example of makers investing what writer and Internet theorist Clay Shirky calls “cognitive surplus.” That degree of engagement has built Wikipedia and released masses of valuable open source software. It has shown itself to have huge potential. Indeed, the hardware development process apparent in Public Lab borrows a lot from the open source software development movement. In early stages, open source software projects tend to be unusable and unreliable, only to evolve into excellent, widely used products. Participants in Public Lab’s projects express the hope that their hardware projects will follow the same path.
  • Public Lab’s work crosses into many fields attracting skepticism. Public Lab, being a complex, multifaceted organization, does things that look a bit like science and engineering (experiments and data collection) and a bit like journalism (producing and distributing politically potent information). Practitioners in those fields apply their own traditional values like accuracy and verifiability when assessing Public Lab’s activity. While Public Lab’s people recognize those values, thus far they appear to have prioritized accessibility and community engagement.
  • Open source hardware development is behind open software. When compared to the breadth, depth, and maturity of open source software, open source hardware appears quite infant. While the tools and ecosystem supporting open source hardware have developed a great deal in recent years (the case study examining the WNYC Cicada Tracker goes into more detail on the subject), some headwinds remain. Whereas a new open source software build is generally free (apart from minimal computing resources), whenever an opensource hardware project moves from design to build, even as a test, it requires physical resources. The tools always have a hardware cost, and even though those costs have dropped, they remain significantly above zero. Lessons for the Industry
  • Engagement needs benefit. Public Lab has, in just four years, built a very large community of people, some of whom have contributed large amounts of effort for no payment. Crucial to that degree of engagement is that individuals’ participation delivers them some kind of benefit—learning new skills, being heard in a debate, affecting civic processes, uncovering information about their local environment, etc. Public Lab’s practice of eschewing top-down control in favor of self-determination undoubtedly helps with that. Many media organizations include engagement in their objectives, but few are able to release as much control as has Public Lab and are often focused on the benefit the company gets from participation, as opposed to the community.
  • As a journalist, beware that official data may be contestable. The processes that produce official data can be complex, opaque, and ridden with errors. In the case of data describing where oil from the Deepwater Horizon made landfall, the production was contracted and subcontracted. At the first stage of the production chain were people on boats, working in a difficult environment, using a newly implemented process. However, those messy, human production processes are rarely brought to journalists’ attention when the data is supplied.

USA Today—Ghost Factories

Alison Young says she is a junkie for lead smelters. She follows that up with a chuckle, but given her track record, it is probably a fair label. It is also a characteristic that has won her numerous journalism awards and underpins pollution watchdog stories she has reported for four different newspapers over 13 years. Over that time, as the techniques and technology of journalism evolved, Young’s reporting has incorporated more and more environmental data collection. Her ambitious project, USA Today’s “Ghost Factories,” saw reporters take more than 800 field tests of suspect soil and prompted the US Environmental Protection Agency (EPA) and state environmental agencies to launch new cleanup efforts. Young wrote her first report about the toxic remains of a lead smelter while she was working for the Dallas Times Herald in 1989. Despite the EPA’s claim that the site was clean, the soil in the area still held large amounts of lead—a contaminant linked to kidney damage, heart problems, and slow IQ development in children. Twelve years later, working at the Detroit Free Press, Young edited a series of investigations into lead poisoning from gasoline and old industrial plants. A group of reporters, led by Tina Lam, published a story about the contaminated Master Metals factory site just north of Detroit midtown. At the time, in 2003, they outsourced the soil sampling to one of the country’s leading experts in soil contamination, Dr. Howard Mielke.

Around that time, very little attention was being paid to a study published by Dr. William Eckel in a 2001 edition of The American Journal of Public Health. He wrote that up to 430 old metal working factories had been closed and forgotten, potentially leaving behind heavy metal poisons at the sites and in surrounding neighborhoods. The families that moved into the areas had no way of knowing that their children could be playing in unsafe soil. But the study found its way to Young. “I have always done a lot of reporting in the area of public health and environmental health and so one of the first things I tend to do is a search of the scientific literature,” she says. In October of 2009, working in Atlanta, Young again pulled out Eckel’s list of forgotten industrial sites. Just before she left The Atlanta Journal-Constitution she wrote a story about the old Evans Metal Company, a lead factory that had been shut down, sold, and redeveloped with no EPA oversight to ensure the plant’s land and its surrounds were clean. So when Young arrived in USA Today’s investigative unit in November of 2010, she was primed to do a nationwide investigation based on Eckel’s list of old smelter sites. Blake Morrison, the editor of the new investigative team, was a longtime reporter at USA Today and had worked on a series called “Smokestack Nation,” which had included air pollution sampling techniques pioneered by The Houston Chronicle (which are discussed in an earlier case study of this report). From the outset, Young and Morrison anticipated that their investigation would need to include soil testing, but Young expected they’d need to outsource the work to Dr. Howard Mielke, as she had at the Detroit Free Press. However, while Young was in her research phase, reading the academic literature and government reports, the papers’ authors kept referring to a new device: the X-Ray Fluorescence (XRF) analyzer. Handheld XRF analyzers look a bit like a ray gun from a sci-fi cartoon. They shoot x-ray beams into a sample, pumping energy into the atoms. Each substance’s electrons absorb and lose energy at different rates, throwing light back to the XRF analyzer at characteristic wavelengths. An onboard sensor detects the emissions and

compares them to a library of known patterns, logs the data, and shows the results on a small display. The fact that so many different people were using these new tools hinted to Young that the reporters could do some of the data collection themselves, bypassing the huge expense of outsourcing the sampling and increasing the amount of data they could collect, economically speaking. Hiring outside experts to take field samples nationwide would have been incredibly expensive. Typically a newspaper would have to pay consultant rates for multiple scientists to travel to each location, take soil samples, and then do the processing back at their lab. A company called Thermo Fisher made a leading line of XRF analyzers, used in the fields of environmental regulation, resource mining, and industrial production. Although Young hadn’t seen any examples of journalists using them, the tools appeared to be relatively easy to operate. “Blake Morrison was saying, ’Well, what if we could?’ ” says Young. “That’s a very important thing when you’re trying to do something that hadn’t been done before.” Although the upfront purchase cost would have been $41,000 per unit, Morrison was able to negotiate a rental agreement of $2,250 per device, per month. Still, having worked in health and science reporting for a long time, Young knew that their process had to be valid. “If we can’t do it right, we put any journalism we do at risk.” Developing the Methodology Alison Young consulted Dr. Mielke to help her develop their reporting plan. “It’s not enough to have the tools, or the sensors. Just like in database journalism, if you don’t ask the right question, you’re going to get garbage back,” she says. “If you go out into the field and start testing without a valid plan, you’re also going to get garbage.” Young’s project was different from the types of randomized field studies that health researchers like Dr. Mielke normally conducted. His teams were used to looking for problem areas across a whole community. Meanwhile,

health authorities’ studies were often designed to understand the total cleanup costs. In contrast, Young’s project would test her hypothesis that the contamination from old lead factories fell on the soil in particular locations— putting families at risk. Young drafted her sampling plans and sent them to Dr. Mielke, who helped her assess whether her protocol would produce the data to answer her question. The final plans dictated that the reporters take their samples within a mile of the old factory sites, mostly in the direction of the prevailing winds. They reviewed aerial photographs to avoid confounding factors like current industrial plants and locate soil that likely hadn’t been disturbed—mostly bare grass in backyards. The XRF analyzer USA Today used was designed for field sampling. It’s selfcontained, it self-calibrates, and has onboard software to stop users tampering with the data—a feature put in place to meet the standard for regulatory use. But industrial clients also use the devices to check that the steel parts they receive match the chemical specifications they have paid for. Miners use the device to analyze soil. Workers across all these different industrial applications set the units with subtly varying settings. As part of the rental contract, the tool’s manufacturer, Thermo Fisher, set the units to analyze soil for specific pollutants for USA Today. “Originally, we were going to report on arsenic and a variety of other compounds in the soil,” says Young. “For simplicity’s sake [in] explaining to readers, we just stuck with the lead, but we wanted to be able to capture as much data as possible, in case we needed or wanted it later.” USA Today also contracted Thermo Fisher to send training videos and instructors to provide hands-on instruction for Young, her co-reporter Peter Eisler, and a backup reporter in case Young or Eisler got sick.

The Reporting When Alison Young and Peter Eisler got into the field, they were carrying the XRF analyzers, their field notes, cameras, and microphones to produce the video they wanted. The reporters also had legal releases and documents for the residents living on the property where they took most of their samples. The releases informed the residents that their exact street addresses wouldn’t be published and that their names wouldn’t be used unless they specifically agreed. Dr. Mielke had suggested USA Today write up information sheets to leave behind for the homeowners, who suddenly had to absorb the possibility that the soil in their backyard might be harming their children. The leave-behind information packs gave the residents the results of the tests for their backyard, along with contact details for local environmental agencies, regulators, and some resources to find out more information. In backyards, public schools, and sports grounds the reporters took more than 800 surface samples with the XRF analyzer and collected around 190 additional samples of earth to send to Mielke’s lab at Tulane University for more exhaustive analysis. Young and Eisler were on the road for almost two full months. The expensive lab tests performed at Tulane correlated with the recently trained journalists’ field sample results. The field tests “revealed potentially dangerous lead levels in parts of all 21 neighborhoods examined across 13 states.”1 Although the EPA says that 400 parts per million (p.p.m.) is the upper safe amount of lead in soil where children play, states set their own regulatory limits. For example California’s limit was 80 p.p.m., whereas New York’s was 400 p.p.m. Tests that Young and Eisler performed using the XRF analyzer “showed several neighborhoods had lead levels greater than 2,000 p.p.m., topping 3,400 p.p.m. in Cleveland, Ohio; Portland, Ore.; and Carteret, N.J.” A baseball field in Red Hook, N.Y., showed lead levels of 2,000 p.p.m. When USA Today informed city officials, they shut the park down until the fields could be resurfaced. Six out of 16 tests performed in nearby public housing yards showed levels over 400 p.p.m. Young and Eisler’s story documented neighborhoods in Philadelphia where children were playing in soil with lead concentrations more than double the EPA’s limit. Although blood testing was not part of USA Today’s reporting plan, parents living in a contaminated Philadelphia neighborhood had their child tested, and found a blood lead level of 0.075 milligrams per liter, a concentration “associated with decreased IQ and an increased incidence of ADHD and other issues, medical studies show.”2 In many places where the reporters found dangerous pollution, Young had evidence that the EPA had either gone looking for the closed factories documented on Eckel’s list, but failed to find what USA Today did, or knew there were risks and took no actions. Many state environmental agencies were similarly ineffective. Publishing Although the data from the soil tests was crucial to USA Today’s finished story, the newspaper company’s online publication in April of 2012 took the form of an extensive multimedia presentation. Aside from many thousands of words, it included video segments, photos, embedded government documents, and an index of the sites on Eckel’s list with map overlays showing historical plans of the factories and the soil sample data from the reporters’ testing. USA Today staff hand-positioned its data markers on the map so that the pins didn’t fall directly on houses where residents had elected to keep their addresses private.

Young says that after publication they took phone calls from state regulators, who pointed out that they hadn’t followed the randomized process the regulators used to start enforcement procedures. Young’s response, she says, was that USA Today had produced enough reliable information to show that even though William Eckel’s study had pointed to neighborhoods where old smelter sites posed a high risk of poisoned soil, the regulators didn’t know whether the land was safe and there was enough data to suggest they should find out. Mielke used the term “screening” to describe the utility of USA Today’s data. He says that, as a first pass, it produced reliable enough evidence to suggest environmental regulators should invest the extra resources to procure legally sound data in preparation for a cleanup. After USA Today’s first publication, the journalists continued to report on the series’ impact. In Carteret, N.J., the local government reached an agreement with the owner of a closed factory, who would pay $1 million to assess and clean up the surrounding area. In Portland, Ore., the EPA agreed to pay for cleanups at five homes the reporters had tested, and committed to testing at a sixth. The EPA inspector general, who was responsible for auditing the agency’s operations, prioritized a review of its performance in reducing health risks from old lead smelters. U.S. senators from New Jersey, Ohio, Pennsylvania, Rhode Island, Oregon, and Minnesota pressured the EPA to take immediate action, review all the sites, and set priorities for remediation. In November of 2012 the EPA announced it would reexamine 460 former lead factory sites. However, Alison Young says the testing and cleanup actions from national and state environmental agencies were uneven. When she asked state regulators, before and after USA Today published in April, why they hadn’t tested, the answers were always about lack of funding: “Literally, one of them said to me, ’because if we do, and if we find something, then we have to do something, and we don’t have the money.’ ”

Distinctive and Notable

  • Dr. Eckel’s paper had been available for 11 years when USA Today published. The newsworthiness of Dr. Eckel’s paper identifying 430 forgotten, potentially toxic smelter sites evolved and grew over the 11 years between its original publication in 2001 and USA Today’s report in 2012. Initially, the story was simply that these sites existed and environmental agencies likely needed to take action. Over time, it shifted to outlining that these sites, probably toxic, were now known, and environmental agencies were continuing to take no action even though public health was probably at risk. That lack of action became less excusable as time passed. USA Today’s story demonstrated that the risks were real and relatively easy to confirm.
  • The development of XRF analysers opened up new reporting opportunities. Although environmental testing was a known investigative technique before this story was reported, the advent of a rentable electronic, self-contained devices that journalists could learn to operate meant USA Today could take hundreds more samples than was previously economically viable.
  • The journalistic hypothesis required a different sampling approach. The differences between Alison Young’s goals and other applications of soil testing came into play at two main points in USA Today’s project. When she was designing her testing process, she was trying to verify whether a particular pollution source was endangering particular residents. She therefore could not use a template from a health researcher’s project where the goal was to get a representative sample from a large area. When USA Today published, according to Young, experts who worked in environmental agencies said that her process was not the same one they would use to plan and estimate the cost of a cleanup and suggested her story was therefore invalid. That tension appears to be the result of practitioners from two different fields using similar processes for different ends.
  • Having an adventurous editor helped. Young suggested a reporting process incorporating XRF analysers, which journalists had not used before. That entailed risks. However, her editor, Blake Morrison, had previously used sampling while reporting on his own large environmental projects with some success, and supported Young’s concept.
  • Health authorities, experts, and states do not agree on standards. Part of the impact of “Ghost Factories” came from its hard, quantified data. The reporters were able to produce numbers that were ripe for comparison. However, there was room for debate about how to compare their data. Although some of their soil samples had lead concentrations above 2,000 p.p.m., which were clearly unsafe for children’s play areas, many samples had lead concentrations in the hundreds of parts per million. USA Today emphasized the EPA’s 400 p.p.m. in its presentation, but also noted that California set its limit at 80 p.p.m. In later stories, Young wrote that many experts believed that a 400 p.p.m. limit for children’s play areas was too high. Lessons for the Industry
  • Without good technique, the technology won’t help. When the Tow Center spoke to her, Alison Young repeatedly emphasized that the newly available XRF analysers could have easily produced useless data. Alternatively, the data they did produce could have been interpreted incorrectly. Two factors gave strength to her findings. First, she had a clear hypothesis that could feasibly be tested using the technology. Second, the reporters developed a testing protocol, with input and review by an expert in the field who helped ensure it had rigor and validity.
  • Initial research reduces the risk of finding nothing. By the time Young and Morrison made their pitch to USA Today’s senior editors, asking them to approve an extensive reporting process and a large cash spend, they had filed FOIAs for EPA documents, knew the dominant wind directions that would influence the most likely problem spots, had read the scientific literature—all on top of Young’s long experience covering similar stories. This offered their editors confidence that there was probably a story to be reported, and that evidence could be collected to write it.
  • Make privacy a high priority. Many of the residents who allowed USA Today to sample the dirt in their yards wanted to remain anonymous. The journalists needed to respect their privacy while still showing relevant results on a map. So, USA Today staff, having initially placed the map pins based on residents’ street addresses in an automated fashion, had to move the pins manually onto nearby median strips and public space.
  • Prepare leave-behinds to help your sources understand your findings. Dr. Howard Mielke, a research professor with long experience doing environmental testing in residential areas, advised Young to prepare for situations in which the people whose yards she tested would want help understanding the data she had just produced. Understandably, he anticipated that they would probably be overwhelmed: This was important information with serious implications, and in contrast to the reporters who had been immersed in the data for many months, there was no guarantee that the residents would be familiar with lead pollution and its health effects. However, while the reporters could supply the specific data they found and some basic background, for more information or to take action the sheets pointed the residents toward health and environmental agencies.
  • Be prepared for non-technical reporting obstacles, like language barriers. Outside of technical sensor use, Young says she wishes she had developed a plan to deal with the language barrier her team encountered. While they had prepared for all kinds of testing, their inability to speak languages other than English in some neighborhoods limited the testing they could do.

Sun Sentinel – Above the Law

For police officers in many Floridian cities, a longstanding perk of their jobs is a take-home car—even relatively junior officers get one. The benefit offsets the job’s low public sector wages and the dangers of duty, police unions argue, while attracting recruits and keeping them in the force.1 But before Sally Kestin and John Maines from the Sun Sentinel in Fort Lauderdale wrote their Pulitzer Prize-winning stories about speeding police officers, Florida’s cops were driving thousands and thousands of trips at more than 30 m.p.h. above the speed limit; breaking the law, causing fatal accidents, and escaping heavy penalties. The reporters’ work eventually forced the police departments of Miami, Fort Lauderdale, and eight other state police departments to launch internal affairs investigations cracking down on speeding cops. Follow-up reporting showed that the series shifted the police culture from one of blatant disregard to safer compliance. What Kestin and Maine’s reporting produced is a story in which sensor data revealed a secret culture of police disregard. While outliers and special cases were crucial to give the narrative bite, they weren’t at its heart. Though the duo didn’t necessarily set out to combine sensors with data analytics and investigative reporting, these are their story’s driving forces.

Through 2010 and 2011, Sally Kestin says Floridians were well aware that their off-duty police were smashing speed limits along the expressways. One of Kestin’s readers would call to bemoan that he always saw cops “speeding like crazy” on his daily commute. When Fausto Lopez, a Miami police officer, sped past trooper Donna Jane Watts’ car on October 11, 2011, he was clocked driving his marked police car at 120 m.p.h. in a 65 m.p.h. zone. Trooper Watts accelerated to chase Lopez and pull him over but the officer wouldn’t stop. The video from Watts’ dash-mounted camera shows Lopez speeding away for eight minutes before finally acknowledging Watts’ commands. When the story and the video become public, newspaper editorials and social media erupted. But the problem remained: how to debunk the Floridian police departments’ standard defense that Lopez—and perhaps a couple of other cops— were merely a few bad apples. The Sun Sentinel wanted to show that while Lopez was certainly an extreme example, he wasn’t such a radical outlier. Mid-November of 2011 found John Maines and Sally Kestin standing in the rain by an overpass above Florida’s Turnpike, pointing a speed gun they’d just bought at cars as they whizzed through the narrow strip of suburbia that separates the Everglades from the beaches above Fort Lauderdale. In the most optimistic scenario, it was conceivable that the reporters would be able to collect a few hundred speed readings, enough data to expand their own picture of speeding cops from just a handful to the low hundreds. But, as investigative reporters, they quickly understood that their data would be highly contestable: Police and other skeptics could argue convincingly that the equipment wasn’t reliable or that Kestin and Maines couldn’t be trusted to operate it properly. The data would not be reproducible. The two journalists asked police departments for data from the GPS units that tracked the location of police-owned cars, but the departments said delivering that information constituted a security risk to officers who parked their cars in their home driveways.

Once again, it was a reader who called to complain about police speeding that helped Kestin move the investigation along. The reader mentioned the tollgate data from the expressway. “’Well, you know SunPass has these dates and times,”’ Kestin reports the reader saying. “So that’s how it originally came up—in a conversation with him,” she says. The data reporting for this project was only possible because Florida had built connected, physical sensing systems that could accommodate the complexities of the bureaucracy’s financial transactions. Florida police were not only entitled to take-home cars, but those cars were allowed to travel for free on the state’s tollways. However, the highly automated world of SunPass still needed to know that the car passing through the tollgate was legitimately not paying. Theoretically, the police cars could pass through, be photographed, and SunPass employees could check the number plate against a registration database, conclude that it was a police-owned car, and drop the fee. But that would be a wasteful process. So the police attached a standard SunPass RFID transponder to each state-owned car. In SunPass’ back-office systems, all of the police’s transponder IDs were linked to nonrevenue accounts. The tollgates would ping the transponders, register their passage at the appropriate time, and send the data back to sit in a central repository, but send no charge to the drivers. The ingredients of a proof, therefore, existed—at least theoretically: Each car’s transponder had a unique identity. For each trip, that identity was recorded a minimum of two time stamps, at knowable locations. The reporters could take the time difference between the waived tolls at two locations, divide it by the distance between those locations, and calculate an average speed. But to lock it down, Kestin and Maines needed to be sure that the time stamps were accurate and calibrated, that the gates’ locations were known and that all the gates registered transponders at the same place within a few yards of the hardware. Also, they had to actually get the data, which is where Florida’s relatively journalist-friendly culture of public disclosure

helped, along with some luck. In contrast with much modern infrastructure around the world, the government owns the tollways. Kestin and Maines could mount an argument that the data was public record, not private business information. Kestin and Maines contacted SunPass with their request, instead of the individual police departments. The first time around, SunPass refused to release the data, quoting the Florida laws that said personally identifiable information was exempt from Freedom of Information Act requests. The journalists argued that there was not personal information in what they sought, just transponder numbers assigned to a city agency. Maines says, “SunPass agreed after only a few days,” and “started providing us with information,” handing over 250MB-worth of Excel files containing 1.1 million rows. When John Maines discovered that every gate in the SunPass network was synchronized to the National Atomic Clock in Washington, D.C., he knew they had accurate, calibrated time stamps. Getting the locations was harder since they were recorded in the data as plaza names, with no lookup for latitudes and longitudes, nor any mention of the length of road between

one gate and the next. It took Kestin and Maines shoe leather, (or at least tire rubber) and a few trials and errors, to produce those missing pieces of the data. Kestin and Maines’ first idea for a tool to measure the distances between gates was the odometer in their cars. They would simply drive the roads and mark down the numbers as they passed under each gate. But when they tested the process, they found out that Kestin’s Lexus and Maines’ Mini Cooper registered slightly different distances, even over trips as short as 1.6 miles. Worn tires, variation between tire brands, and the general inconsistencies between cars introduced too much uncertainty. Their next move was to convince Garmin, a technology company that designs, manufactures, and markets GPS navigation equipment, to loan them an Echo200 GPS unit—a $130 tool that would give them accurate, reproducible locations and distances. Over the next weeks Kestin and Maines drove almost every stretch of tollway in Florida. In many cases, they had to drive each roadway in opposite directions to cater for the variations in on and off ramps. Through it all, the Garmin returned readings accurate and consistent to within 1/100 of a mile. (John Maines says that when he ran spot-checks with Google Maps’ driving routes tool they returned distances consistent with what they’d measured—at least precise enough to calculate broken speed limits. But they didn’t know Google would be accurate until they’d driven the routes themselves.) Finally, all the raw data was in place. John Maines’ next task was to write the algorithm in Microsoft Access to show all the trips that broke speed limits. “When you’re doing calculations on tens of thousands of numbers and you get a formula wrong, everything is going to be wrong. So if you screw up it could be the end of your career,” he says. “The first time I did it, it took me a few weeks. It would throw out all kinds of errors.” Meanwhile, Sally Kestin was reporting out the rest of the story, talking to people who’d been involved in accidents with police, and asking why, when those accidents happened, the police escaped heavy punishments.

Nonetheless, the foundation of the story rested on processing the data, using methods that were new to Maines. “If I had a chance to go back in time,” says Maines, “I would have told myself how to do this. But I was very proud to have been able to produce it, and in the end [the results were] very good, very accurate.” Kestin and Maines’ reporting work seemed to show that between October 2010 and November 2011 Florida police drove their cars more than 6,000 times, not just above the speed limit, but in excess of 90 m.p.h. According to Kestin and Maines’ process, the data held hard evidence of what Floridians had suspected and complained about: The state’s police forces were ignoring their officers’ disregard for the law and, despite deaths and injuries, they had allowed a culture of reckless driving to grow in their departments. Despite Kestin and Maines’ methodical approach, their editors still pushed them to prove their process was sound, ideally by consulting someone who’d done it before. John Maines was assigned to find someone. “Sally knew of a traffic engineer, so I called him and he said, ’I can’t really talk to you about it, because I have cases like that in front of the state attorney’s office.’ ” It turned out that the engineer had taken the dash-cam video from highway patrol trooper Donna Watts, who was chasing Lopez, and recorded the time at which the chased car went under each overpass. That process was close enough to theirs to give them some confidence. Kestin says her traffic engineer source calculated Fausto Lopez’ speed using the SunPass records too—the same way they had. “We took all of our findings to each department we were going to include in the story,” says Kestin. “We took it to them well in advance of publication to give them a chance to vet it, to make sure that we had not overlooked anything, to make sure our methodology was rock solid. We said, ’Look, this is what we got. You tell us—is this what it appears to be?’ All of them opened internal affairs investigations at that point, before we even went to press.”

Publication Knowing that they had gotten the story and that it was right, Kestin and Maines were left with the decisions about what to include and the best way of communicating the facts. Ultimately, they opted to only publish transponder records for the police who had been driving above 90 m.p.h. on three or more occasions—more than 25 m.p.h. over the Florida speed limit. They also needed to choose whether to de-anonymize the data, which would raise questions of privacy and surveillance. Pragmatism contributed to their decision not to: “We could have made public records requests with all 16 police agencies and asked for a list of all their transponders and who they’re assigned to,” says Kestin, “but we knew that they wouldn’t want to give that over. Even though this is the great open records state, we figured that could take another month or two months and it would be a fight.” But above all else, Kestin was more interested in showing the police culture rather than naming individuals. “I would have hated to name somebody and blasted their name, their excessive speeds, all over the paper and then have the guy be cleared two months later. I think we handled it exactly right. I think the power of this was that never before had these police departments been confronted with the sheer proof of how widespread this problem was, and how dangerous it was.” Fausto Lopez was the exception. In October of 2011, when the video of him speeding broke, newspapers and TV stations across the United States had already broadcast his name. Extreme case or not, his was the recognized face of reckless driving.

Figure 1: Sun Sentinel graphic With such a data-heavy project, there was an obvious opportunity to use visualization in storytelling. “I originally did the graph in Excel,” says Maines. “My first thought was, ’Let me show his average speed for a month,’ and that looked okay, then ’Let me show his average speed for a week,’ that looks even better, then I said ’Let me show his average speed every single time he traveled the roads.’ That was the origin of that graph with all the hairlines, and I took it back to the graphics reporter and she put it all together. Everybody loved that one.” The Sun Sentinel went on to win the 2013 Pulitzer Prize for public service reporting. The jury praised “its well documented investigation of off-duty police officers who recklessly speed and endanger the lives of citizens, leading to disciplinary action and other steps to curtail a deadly hazard.” In the months after the Sun Sentinel published the story, nine Florida police departments took disciplinary action against their officers who were routinely speeding. The vast bulk of penalized officers—numbering 130—came

from the Florida Highway Patrol, Miami-Dade Police, and Miami Police. SunPass records showed that the rate of speeding police dropped by 84 percent after the Sun Sentinel’s first investigation. As Kestin and Maines wrote in their follow-up story, “The trend is dramatic and definite: South Florida cops have slowed down big time.” Distinctive and Notable

  • Investigative work quantified what was previously a perception. Before the Sun Sentinel did its investigation, the case against Florida’s police was anecdotal and unsystematic. By accessing accurate sensor data and interpreting it correctly, the journalists produced clear, specific recordings of events occurring in the physical world that could be objectively compared to legal standards.
  • Data was a particularly suitable reporting tool for this story; confirming a widespread practice. A key point that Kestin and Maines succeeded in making was that speeding was widespread. While journalism has historically used individual examples to illustrate a problem, often with accuracy and great effect, the rise of data and accompanying visualizations have helped journalists better describe widespread phenomena in a memorable and resonant way.
  • A reader sparked the idea for the reporting technique. Kestin’s relationship and interactions with a reader were crucial to the reporters’ development of their innovative reporting process. It was a concerned and motivated community member who suggested that SunPass recorded car movements and might be a viable data source.
  • The data processing was done in Excel and Access. At the 2014 NICAR conference (the premier data journalism meeting), the schedule only mentions Access three times, and two of those mentions are pejorative. Although R and Python are the chosen tools of many new journalists working with data, in this highly successful example John Maines used Microsoft Access as his data processing workhorse.
  • The first sensing and verification plans did not work. Kestin and Maine originally tried to use a radar speed gun to collect data before concluding that it wouldn’t work. Likewise, their first idea for measuring the distances between tollgates was to drive the roads and take odometer readings. In both cases, they did not begin with the expertise to know the techniques wouldn’t work. However, when their research revealed a problem, they changed their approach.
  • The availability of the data relied on SunPass being a government-owned company. The highways and toll-collecting infrastructure in Florida are publicly owned and operated. That is not the case for all civic infrastructure. In the words of one influential study into public and private partnerships, “Governments worldwide have sought to increase the involvement of the private sector in the delivery of public services.”2 There are privately run toll roads in states including Indiana, California, Texas, and Illinois in the United States. Overseas private toll roads exist in the United Kingdom, Australia, and Spain. It is not clear that Freedom of Information Act requests would succeed if they were filed asking for information from tollgates on those roads.
  • The data contained errors—which were valid. A standard phase in journalists’ data synthesis is a “cleaning” one. For data that has been produced by manual, human processes, this often includes merging alternate notations of the same concept; misspellings of names, abbreviations, and different labels for the same object. John Maines, when describing the SunPass data, said it was “clean,” meaning it was consistently formed and valid. Therefore, the data passed through his formulas without breaking. However, some of the data recorded in the original SunPass files, while valid, produced results which, when given an eyeball review, were implausible: speeds of 6,000 m.p.h., for example. While automated error checks could certainly be written to replicate Maine’s human assessment in the case of a calculated speed 100 times the limit, this is a different class of wrongness, which may not be as easily detected as the kind of errors produced by data from manual processes.
  • Two people, three months. The vast bulk of the work for this Pulitzer Prize-winning story was done by only two people—Sally Kestin and John Maines—although editors provided guidance and, in the later stages, editorial assistants, graphics, and video staff contributed. As Kestin says, “This project, from start to finish, only took us three months. In the world of investigative journalism, that’s not a long time. So it’s possible to do really high impact stories without having the big publications’ luxury of spending a year or two and really make a difference in your community.” Lessons for the Industry
  • Acquire a granular understanding of the sensing mechanism. Kestin and Maines made their story bulletproof by, among other things, gaining a fundamental understanding of exactly how the tollgate sensing worked. They found out how the tollgate clocks were synchronized. They investigated exactly where, relative to the tollgate overhead beams, the RFID tags on cars were detected. Beyond the mechanics, they understood the administrative processes behind the data collection, all of which increased their confidence in making sound conclusions.
  • Surround the data with human context. Although data provided the conclusive evidence of a widespread problem, the reporters added touching and powerful stories of individual people who’d been killed by police driving speeding cars.
  • Verify your process with authorities. Despite the Sun Sentinel’s internal checks and balances, review by editors, and cross-checking with sources, Kestin and Maines still showed their data and their process to the police departments featured in the story.

WNYC – Cicada Tracker

Good radio stations, perhaps more than any other traditional form of media, thrive on becoming attuned to their communities. Radio hosts like to say that talkback radio relied on user-generated content long before online comment threads and photo uploads became de rigueur in the early 2000s. It’s exactly this culture of participation that inspired the WNYC “Cicada Tracker” project from the outset. It’s not immediately obvious that listeners would turn to a radio station to learn about electronic prototyping, nor that RadioShack would play a part in predicting a once-every-seventeen-year event in an insect lifecycle. Since WNYC’s data editor, John Keefe, attended one of Make: magazine’s Maker Faire events and took up a hobby of tinkering with electronics, he was looking for a way to combine WNYC’s participation culture with the emergent maker movement. When he launched the “Cicada Tracker” project, perhaps he might have predicted that folk imbued with New York’s entrepreneurial spirit would also find a way to make the project their own. In the early phase, when Keefe and his team were casting about for ideas, their rough concept involved distributing some kind of custom electronic sensor around a neighborhood, recording some data, and having it sent back to WNYC. And although they knew their community was interested in electronics they didn’t have a hook to pull people in. In the winter months of early 2013, a little story lead found its way to the WNYC newsroom. Every seventeen years on the northeastern seaboard of the United States, when winter turns to spring and the temperature starts to rise, the local variety of cicadas (called magicicada) crawl out of the little holes where they’ve been hibernating. Unlike most cicada genus, all magicicadas develop in synchronization, reaching adulthood in the same year. Through the summer after they’ve emerged and dried out their wings, the males’ distinctive buzz calls out for mating. For residents near cicada breeding grounds (rarely the inner city) the cicada months can mean weeks of a constant background drone. A good predictor of the cicadas’ arrival, Keefe read, was soil temperature. When magicadas hit their adult year they emerge when dirt eight inches below ground reaches 64 degrees Fahrenheit. And that is the kind of natural phenomena that’s fairly easy to sense electronically. Many popular starterelectronics kits include tutorials and parts necessary to build an electronic temperature sensor in just a couple of hours. The sensors could be placed in the communities’ backyards, producing data to predict when the cicadas would come out. WNYC would need to add a few customizations, but all the sensors could be made with easily accessible parts. The ideal public face broadcast for the project was NPR’s hugely popular science show, Radiolab. In addition to the program’s loyal, enthusiastic community, it had a grant from the National Science Foundation to deliver informal science education. The WNYC data news team initially planned to build up to a dozen sensor nodes and have the data sent back over wireless connections. They wanted to keep the process simple—easy enough for an amateur to replicate—so during a hack day, they made a parts-run to RadioShack and spent about $200 on an Arduino electronics prototyping kit and a set of accessories. They then prototyped the wireless version of the sensor, placing a thermistor at the end of a short length of wooden dowel with wires running back to a little microchip. They wrapped a plastic baggie around the end of the dowel that held the temperature sensor to protect it from moisture when it was stuck eight inches into the ground—or in this case, a nearby potted plant. At the hack day, the temperature readout came up on a web page. Later, the team opted for a manual data collection method, using nine LEDs for the temperature readout, which was far less fussy than trying to set up a wireless or cellular link. Altogether, the sensor design that WNYC published online required parts costing about $80 and could be assembled in 29 steps, each illustrated with a picture and a short piece of text. Consequently, people in the WNYC community actually spent the money, invested the time to build the sensor, and started to send in their temperature readings. Radiolab presenters Jad Abumrad and Robert Krulwich talked about the project on air and promoted hack days for listeners to learn how to build their own sensors. The team wrote stories online and recorded segments with a professor of evolutionary biology at the University of Connecticut, Chris Simon, who’d be collecting the data for her research. And then, the community went from following the instructions to rewriting the playbook. In a couple of rooms above 14th street in New York, not far west of Union Square, there’s a small maker-space set up by a group called Hack Manhattan. An indulgent landlord rents the space to them at a heavy discount, lets them fill it with a mess of half-used mechanical parts, electronic works-in-progress, a home-brew beer setup, multiple 3D printers, and a few benches where members come in to work on whatever kind of craft project piques their interest. Guan Yuan was one such member, who worked on occasional electronics projects between studying business at NYU, tweeting wittily and prolifically, while blogging occasionally. He’d also been to a class run by NYC Resister, a hacker collective based in Brooklyn with the slogan “We learn, share, and make things.” They’d started to teach him how to take an Arduino prototype and pare it back to its essential components; making it cheaper to build, more efficient to run, and devoid of any wasted resources.

The Arduino electronics platform was released in 2005 by an Italian interaction design professor named Massimo Banzi who wanted a tool for his students to learn electronic hardware development. At the center of an Arduino kit is a cheap microprocessor that runs code recognizable in the c++ family. The microprocessor is attached to a circuit that connects power, a numerous and varied set of communication pins, and a USB port that can deliver data to a Mac or PC computer for programming. A store-sold kit normally includes a bread-board, which developers can use to wire-in sensors, data loggers, wireless communications, lights, or any number of different interfaces they choose—all without ever powering up a soldering iron. Arduino’s open source hardware and software affords accessibility, low price, and lots of flexibility. These attributes have made it wildly successful. By April of 2013 customers had bought more than 700,000 official boards and the founders estimate that for every official board there is a (legal) clone running. However, Arduino’s design and intent come with tradeoffs. If a developer is only planning to use the platform for a single application, he or she won’t need the easy re-programmability that comes with plug-and-play computer connections. Likewise, there will probably be far too many communication pins and the breadboard is surplus to requirements. Through Twitter, Guan Yang heard about WNYC’s cicada project, but could easily see ways that the original electronics design could be re-worked to bring the hardware cost way down. Thus, he intended to sell specialized cicada tracker kits at a Hack Manhattan’s maker day. Part of his plan improved the design and leveraged the remarkable economics of the international electronics industry. The key to reducing cost in his new version was moving away from the Arduino-branded and bundled parts, and buying only the minimum amount of components from a cheaper distributor—instead of using the full starter Arduino kit RadioShack sells at a considerable markup over the parts’ cost. Yang’s kit avoided the use of a breadboard, replacing the flexible

circuit board with a printed circuit board (PCB) and swapping out the overpowered brain of WNYC’s design, an ATmega328P microprocessor, for its slightly cheaper variant, the ATmega168P. He designed the new circuit board using KiCad, a program that’s widely used by the open source hardware community, and sent his design files (known as gerbers) to Smart Prototyping, one of the cheap Chinese board houses that accepts orders for short production runs of electronic parts. The factories in Shenzhen, China, printed the boards, and a couple of weeks later—for a total cash cost of $87.86 (of which $22.80 was shipping)—50 circuit boards arrived in New York via mail; the price per board was just $1.75 and the microcontroller cost about $2.40. So just by replacing the “Cicada Tracker’s” Arduino core with his custom PCB and the microprocessor, Yang reduced a $30 parts cost to $4.15. The other necessary components— the thermistor, LEDs, batteries, and wiring—cost only a few more dollars meaning Hack Manhattan could sell a whole kit, complete with LEDs and a temperature sensor for $20, which still included a good profit.1 Yang says he put a total of 11 hours into the project. While hack day attendees did buy some of the kits with the PCB, a more popular option was an alternate kit the Hack Manhattan members put together (also for $20, but with a lower profit margin), which had a breadboard instead of the PCB, giving buyers the ability to reuse more of the parts for other projects. From WNYC’s first prototype to Yang’s streamlined model, the component cost of the cicada tracker had dropped 90 percent. WNYC designed its own cheaper breadboard version based on Yang’s design and gave them away at public events to listeners and institutions from along the northern East Coast.

As the cicada season progressed, from anticipation, through arrival, into a noisy summer, the WNYC data team opened the doors to reports of cicada sounds and sightings. John Keefe says that by the end of the project, contributors had sent in more than 1,750 temperature readings from 800 different locations. When the cicadas started to emerge, contributors filed another 4,300 reports recording where and when they’d seen new cicadas, aside from any temperature data. Those are big participation numbers for any kind of journalism project, but they’re even more impressive considering the fact that they demanded people first learn about electronics, assemble a kit, stick a temperature sensor into the soil, read the temperature from a pattern of LEDs, and submit the reading using an online tool. As Keefe says, “From an engagement standpoint in journalism, that’s kind of amazing.” Distinctive and Notable

  • An archetype for a particular model of using sensors in journalism. The WNYC “Cicada Tracker” was probably the first time a major media organization had so explicitly combined the activity of electronics and sensor prototyping into a journalism project. They placed sensing at the forefront of the participatory experience. This flavor of media sensor use was distinctive, clearly innovative, and pointed at a new realm of worthwhile experimentation and approaches.
  • Participation and radio go together like sensors and Arduino. WNYC’s efforts to train its community to participate in data projects had a head start. Radio, as a medium, needs a variety of voices to be sonically interesting, and to reflect the identities in a station’s listenership. Talkback is one of the main ways radio stations achieve that (it’s also relatively cheap). An early WNYC data project asked New York listeners to send in their neighborhood’s prices for milk and a six pack of beer. Through running projects like these the online teams of WNYC have developed an instinct for how to evolve their participatory culture from radio to digital.
  • This shows an evolution in digital participation. From mainstream media audiences, we are used to seeing participation in the following forms:2 (Listed here in rough, arguable, order of user effort) 1) Clicking “like” or recommendation buttons, 2) distribution on social media and email, and 3) contributing to comment threads, call-ins, writing letters or emails, photo and video contributions. In this case, the WNYC’s community engaged by building, and then redesigning, electronics. Precedents exist, of course: Open source software and hardware communities flourish online, although legacy media companies’ audience-facing operations rarely interact with them. Make: magazine is built on, and for, such people—and puts them under an umbrella with the craft communities who cook, garden, sew, and knit. In Sydney, Australia, the ABC 702 local radio station hosts a yearly “knit-in.” In the preceding months, its community designs, then builds (knits) garments and toys. That may not be electronics, but it’s still maker participation. Nonetheless, WNYC was the first mainstream media organization we know of to successfully encourage a non-trivial number of people to build electronics and contribute sensed data.
  • In this case, the accuracy of the data wasn’t actually central to the project. Even beyond the question of whether WNYC’s community temperature observations were accurate, one might ask whether the optimum strategy for predicting the cicadas’ arrival was to deploy home-built sensors; especially while having minimal control over their geographic distribution, or the consistency of their operation once in the field. John Keefe has acknowledged that more accurate predictions could certainly be made based on official air temperature readings and soil-type models. However, that question mistakes the benefits of the “Cicada Tracker” project, whose goals included informal science education. Keefe argues that many of the people who built the sensors were using Arduino for the first time, and were thereby exposed to electronics prototyping—a growing movement. They learned about the relationship between cicada life cycles and their natural surrounds. They also enjoyed the benefit of participating in a communal activity. The data may or may not have been accurate or useful, but it is inadequate to emphasize that goal to the exclusion of the project’s other outcomes.
  • The conditions were perfect for a price drop from $200 to about $12. WNYC spent $200 on a kit of parts, many of which were superfluous, to build its first version of the cicada tracker. A few weeks later, Guan Yang acquired the parts for about $12. But consider the confluence of events and ecosystems that reduced the cost so much: The radio station operates in New York, a city dense with creative technologists. The project caught the interest of a Twitter community who discussed and distributed the idea. A person who was on Twitter and was interested in electronics had also attended a course focused on just this problem space. The evolution of the Arduino platform made the project conceivable. Arduino itself is an artifact of and an accelerant for the cheap electronics ecosystem that makes it possible to have 50 totally new and customized circuit boards delivered from a factory in China to a door in Manhattan in less than two weeks for less than $100. The implications of this are outlined below.

Lessons for the Industry

  • Motivated and engaged people will contribute a lot of effort. Audiences contribute no reaction at all to plenty of media output, as evidenced by the vast fields of published news stories and blog posts languishing with unpopulated comment spaces and zero “likes.” In this case, however, people in WNYC’s community bought parts, assembled them through 29 steps, wrote code, took sensor readings, and manually contributed them via an online form. One person (operating within his own community who contributed help), went even further by refining WNYC’s designs and producing a newly improved version. This contribution of time, intellect, and work may hint at untapped potential, especially if a project has avenues for entrepreneurial and creative community members to get involved.
  • Use publicity and ease-of-use techniques to engender motivation and engagement. The “Cicada Tracker” project pulled some good moves: Although the project originated at WNYC’s data desk, Radiolab heard about it, reached out and came on board as its public face. Radiolab’s audience is interested in nature science and the hosts have a strong relationship with their community. WNYC designer Louise Ma made the instructions extremely easy to follow, reflecting the project’s overall approach of very clear communication, assuming no prior knowledge but without condescension. The tone of the project combined learning and fun. One WNYC event was held at a beer brewery where attendees could make sensors and meet Radiolab host Jad Abumrad.
  • Be open to the community. The unanticipated and unasked-for innovation by the Hack Manhattan members significantly expanded the project’s reach.
  • Realize the implications of cheapening technology. WNYC’s “Cicada Tracker” project appears to demonstrate the economic feasibility of journalism projects producing and using customized electronics. Achieving a parts cost of less than $20 seems to open up the potential for media companies to deploy large numbers of sensors (or have lots of their community build and deploy sensors) designed to produce data for just one story. While this was the case for this particular project, there were a number of distinct factors that were not universal. Guan Yang, who has a set of electronics design and production skills that are currently rare, contributed his time at no cost. The “Cicada Tracker” also comprised a basic set of hardware: the low-end thermistor was a very cheap sensor; it didn’t need a screen display, data storage or networking; and there was no real miniaturization nor need to make an appealing form-factor. If those factors are not in place, customized hardware development can get very expensive, very quickly. At a unit price of $100, a project requiring just tens of sensor nodes moves beyond the budgets available to most newsrooms, and above what many individual audience members are willing to spend. Nonetheless, if the field continues, the skills of hardware design and build may become more widely distributed (even into news developer teams), and the ecosystem may produce hardware components that fulfill more journalistic needs.

Planet Money – Planet Money Makes A T-Shirt

The public radio show Planet Money raised more than half a million dollars to make a T-shirt. Its Kickstarter campaign attracted 20,159 backers pledging at least $25 each for a short-sleeved shirt made out of cotton with a printed cartoon squirrel holding a martini glass—and to pay for Planet Money to document the production process “from seed to store.” Aside from the T-shirts themselves, the end result was a series of radio episodes, a Tumblr, and one of the most impressive interactive documentaries of 2013. The online video unobtrusively incorporated footage shot from a high-end cinematography drone flying around a Mississippi cotton farm. Josh Davis was the project’s video director, brought in by executive producer Alex Blumberg. While the farming chapter, which used the drone photography, is the first bit of video that appears in the documentary, the crew didn’t start planning it in detail until after shooting the other chapters in Bangladesh and Colombia. The overall vision for the film had the story split into chapters. It would start in Mississippi, showing the cotton being harvested, then move through Indonesia where the cotton was turned into yarn and fabric. Next the story would split off to Bangladesh where the factory workers made men’s version of the T-shirt, and to Colombia where they made the women’s.

Throughout the international shoot, the crew had looked for ways to collect unconventional shots. Back in the United States, they continued the trend, strapping GoPro cameras to boxes moving down conveyer belts, and rigging another to the corner of a shipping container that transported the shirts by ship, truck, and train. (In fact, one of their unattended traveling GoPros was ripped from the outside of their container somewhere between Jacksonville, Fla., and North Carolina.) In the film’s first chapter, Blumberg and Davis wanted to show a crucial element of the story—the sheer magnitude of the United States’ modern cotton farming industry. The farm near Clarkesdale, Miss., where they filmed stretched over 4,000 acres of land. The owner, Bowen Flowers, sells 13,000 bales of cotton to the garment industry each season, enough for 9.4 million T-shirts. The wildly successful Kickstarter campaign meant that the project was fairly well funded. They had enough money to think big, in Davis’ words. “If there was ever a time to take a helicopter, this would be it.” But that was before Davis hired a freelance director of photography, Eric Helton, who showed the team a Range Rover commercial featuring footage shot from a drone flying in the Utah desert and Colorado mountains. According to Davis, that was the “oh shit moment” for him and Blumberg. Mostly, they’d not been serious about hiring a helicopter, but the quality and style of the advertising footage made the aerial footage seem feasible and desirable. Indeed, Davis’ only concern was that the video might seem too slick, giving the documentary a commercial feel. Davis only considered one company: SkysightRC, who shot the Land Rover commercial. Two brothers, James and Soren Jensen, run the agency out of Colorado, mostly shooting commercials and adventure films in the United States and nature documentaries overseas. After Davis and Helton talked the job through with the Jensen brothers a couple of times, they decided they didn’t need to ask for any other quotes.

SkysightRC’s normal day rate is $3,500, but the figure moves about depending on how many days the shoot covers, and the kit the brothers take along. For the Planet Money job, they used a Cinestar octocopter—an eight-armed craft a meter across with 14-inch rotors spinning at the end of each arm. Fully loaded it weighs about 14 pounds. The base kit costs about $8,000, but buyers who are looking for better results spend much more on parts like controllable anti-vibration camera mounts, plenty of batteries, and live video downlinks. Despite the emergent nature of the field, and the battles around professional civic drone flying, SkysightRC has been able to buy insurance for its business.1 The crew met in Mississippi for the first time on the night of October 6, 2013, with Bowen Flowers’ farm due to start harvesting cotton over the next two days. The four key shots it wanted to use would take advantage of having a small, incredibly maneuverable craft that could give viewers an exhilarating sense of the vast scale of the cotton fields. Drone enthusiasts have learned to recognize the craft’s distinctive moves; transitions between tight shots and aerial angles, swoops around large objects, and shots tracking moving subjects over long distances. The two brothers from SkysightRC work as a team. On this shoot James took responsibility for flying the craft while Soren watched an LCD screen showing live video feed from the onboard Sony FS100 video camera; he controlled the camera angle remotely. In the pink early-morning light the next day, they sent the drone climbing just above the cotton flowers right up to a height beyond the reach of any camera boom. In another shot they flew the drone very closely over the massive crop picker; and in another it’s skimming along the top of the plants.

On the shoot, James and Soren work standing next to each other. In Davis’ words, “They go into the zone. One’s kind of hippy, the other is kind of serious. They’re calling out the shots and the moves to each other.” The final shot they needed from the drone’s capabilities used it as a static camera mount pointing down from high above as the harvesters crossed underneath. In the final two-minute and thirty-two-second video chapter, the editors used about 40 seconds from the drone footage, of which 15 seconds is the establishing shot and 15 seconds is a closing landscape shot. The three remaining short clips are action shots of the cotton and the mechanical pickers doing their work. The drone-shot footage doesn’t call attention to itself; it’s entirely integrated—only enthusiasts can notice the innovative technique used by the crew. When Planet Money’s interactive documentary went online it spread around social media and has attracted 600,000 visitors who stayed on the page for an average of 32 minutes. In the first few months of 2014, the project picked up second place in the National Press Photographers Association award for Best Multimedia Package, a Photography of the Year award from the Missouri School of Journalism, and a Society of News Design award. Distinctive and Notable

  • The drone was not foregrounded. While audience members familiar with drone footage found it easy to identify in Planet Money’s online documentary, general viewers might be unaware of the production team’s innovative approach. The team’s Tumblr, where it published production updates and featured a couple of photos from the Mississippi shoot day, doesn’t draw attention to the drone itself. By contrast, when some other journalistic entities— like the BBC, CNN, and a few non-mainstream media pilots—have made use of drones in the past producers have pushed the technique into the foreground, sparking plenty of attention on social media and commentary on other media sites but perhaps overshadowing the original story.
  • The team sacrificed a GoPro. While the Tow Center’s main motivation for writing about this case was to study its drone use, the production team’s use of GoPro cameras was another expression of the same technological trends that have made drones intriguing as journalistic tools. Put simply, one can now pay far less to buy technology that can do far more. It was therefore economically feasible for the videographers to put cameras capable of capturing excellent HD video in places where camera operators couldn’t reach—and where the cameras could easily get damaged or lost (probably even stolen).
  • All the conditions were in place for innovation. This was a production with a healthy budget. It had far more money than most single-story projects originating from public radio/online. Although it didn’t have a TV-sized budget, neither was it entrenched in television’s conventional ways of shooting aerial video—from a helicopter. Unlike video journalism teams who work on ongoing productions, sometimes falling into habits, the personnel for this project were pulled together anew. Each had recent experience circulating through different professional fields, from agency photojournalism, to magazines, to commercials. Their brief was to be adventurous; they had the resources to match the ambition and the team was fresh.
  • The Jensen brothers tolerate a lot of legal risk in their business. Through the last year, the Tow Center has conversed with a number of media companies that have ambitions to use drones inside the United States for journalism. Our standard interpretation of the legal landscape is that journalistic use of drones is highly risky. Indeed, the FAA claims responsibility for regulating civic drone use in the United States, asserting that journalistic use would be illegal (see the legal and ethical section of this report for more detailed discussion on that topic). However, the Jensens have built a business that draws some of its revenue from drone use inside the United States. They have also been able to buy insurance for domestic use. It should be noted, however, that they are aware of the safety risks, and work to minimize them. Lessons for the Industry
  • Hire the pros… perhaps. Josh Davis, the video director for this project, was adamant that the only way to shoot drone video was to get specialists involved. He says that after witnessing SkysightRC’s sophistication, level of control, and extensive practice he knows without a doubt that producers should hire professionals. However, as the Tow Center’s recent report on the state of online video suggests, the field is rapidly changing.2 It may be that Davis’ recommendation holds true when producers are aiming to make Planet Money’s variety of very high-production value video, but alternative approaches may be valid for applications like breaking news and community contributions.
  • Let the story dictate the use of drones. In the planning process for this production, the team began by establishing the story it wanted to tell. Once Davis and Blumberg understood how crucial it was to convey the incredible expansiveness of the Mississippi cotton farm, they then started to think about how to render that coverage. As a result, the footage serves the story.

Washington Post – ShotSpotter Tracks Gunshots in D.C.

The Washington Post’s coverage of the ShotSpotter system was actually a byproduct of a different story David Fallis was reporting. Guns have been a major topic around Washington, D.C., for many years. Over a ten-year period, the police departments in Washington and nearby Virginia recovered nearly 50,000 of them. The region has a metro population of about six million people. While reporting that story during the spring of 2013, David Fallis talked to a source who told him about a presentation at a police conference that drew on data from a system called ShotSpotter, an array of audio sensors installed around the city. In Washington, D.C., ShotSpotter’s microphones sit on rooftops and near surveillance cameras, monitoring areas for noises that sound like gunfire. When a microphone hears a likely match, it activates a recorder that saves the sounds and sends the data back to the D.C. Metropolitan Police Department’s centralized control system. As data from other ShotSpotter microphones comes in, it gets compared and triangulated to produce a location alert—all within a few seconds. Washington police say the system means they don’t have to rely on neighbors calling 911 to find out about gunshots and get their patrols out to the right places to start investigations. Fallis thought the system itself was underreported, and that an aggregation of those gunshot reports might have its own story to tell.

Fallis couldn’t get the ShotSpotter story to fit into his original piece about gun confiscations, but over the summer and into autumn he began working on the story with two other Post reporters, Dan Keating and Ted Mellnick, with assistance from Andras Petho, an investigative reporter from Hungary who was interning at the Post. In November The Washington Post published their story as a special investigation on the front page of the Sunday paper, matched online with extensive videos, data visualizations, audio, and graphics. Early on, Fallis got in touch with the D.C. police to obtain the ShotSpotter data. Initially they declined to release it unless Fallis filed a formal Freedom of Information Act request. The application went in during April, with Fallis’ main gun confiscations story due to run in May. The D.C. police department conceded that the data was public record and said it was prepared to release it—however, the intent didn’t translate into action. The problem, according to Fallis, was not one of the police department’s will, it was an issue of mechanics. The police department proposed printing out individual gunshot reports on separate pages, a process that would produce reams of paper. In Fallis’ words, the paper parcels would have been “outrageously large, especially in today’s digital world.” Petho called around to other American cities’ police departments that had installed ShotSpotter systems and found that they were all able to export data as Excel spreadsheets with the touch of a button. The reporters also started to establish relationships with people who worked at the ShotSpotter company. In the end, they asked the ShotSpotter staff to help the D.C. Metropolitan Police Department get them the data. David Fallis’ interpretation of the police department’s apparent lack of technical expertise with its own system hinges on the fact that the reporters’ motivations for wanting the data were so different from the department’s. The police use the system in real time; a gunshot triggers an alert in the system, which tells officers where the gunshot came from, down to an area five-feet in diameter. They send a patrol to the spot to start their work and, in many cases, that’s where their use of ShotSpotter ends. Individual reports might be used in court, but for D.C. police, ShotSpotter is a daily operational tool. Fallis was much more interested in analyzing the data for patterns over the years and spatial clusters around particular neighborhoods to show frequent gun use. The only other individual reports that Fallis particularly cared about were those that could be compared to incidents that the Post had written about—and that was mainly so he could try to verify the system’s accuracy. However, in the case of ShotSpotter, the very concept of accuracy is marred by seemingly unavoidable technical nuances. According to the company, fireworks are the greatest source of false positive reports since they produce relatively short, explosive sounds just like gunshots—further complicating data errors around holidays and festivals. ShotSpotter representatives say that in plenty of the cities where the company has installed systems, gun owners also shoot off celebratory gunfire. The Post reported those shots account for around a third of the gunshots ShotSpotter registers in Washington, D.C. Changes to the system over the years further confounded comparisons between data sets. The FBI was the first to fund ShotSpotter in Washington, D.C., responding to the city’s very high homicide rate. The agency first put the microphones out on rooftops in a few areas. Over time, it passed responsibility for the system over to the Metropolitan Police Department as part of the department’s expansion during the administration of Police Chief Charles Ramsey. In 2007 the police department signed a new contract with ShotSpotter, whose acoustic experts came in and adjusted their sensors’ placements, and installed different kinds of microphones in new locations as new bits of funding became available for the initiative. Comparisons between places were also complex. The police department only installed microphones where it suspected there were gun problems and moved them as the situation changed on the ground. There was no record available of exactly when each alteration was made or what was moved each time, which prohibited the reporters from making local comparisons, such as gunshots in Prince George County to those in the District of Columbia. ShotSpotter wouldn’t give Fallis information about what its systems found in other cities. The company also refused to release a complete list of cities that had installed the system. The reporters also considered whether or not there was a privacy and surveillance angle to their story. Each node of the ShotSpotter system is a combination of a microphone and a customized computer that monitors the microphone and turns on the recording when it recognizes a sound matching a gunshot signature. ShotSpotter says that the wave files the system sends back are only a few seconds long. In some cities, ShotSpotter has been integrated with security camera systems, so that when it identifies a gunshot, cameras in the vicinity rotate to point at the calculated origin. However, this has yet to be implemented in Washington, D.C. Ultimately, the reporters concluded that there was not a significant privacy angle to their story. The data that the Metropolitan Police Department released came in an Excel file, with a relatively simple structure: Just over 39,000 rows; one for each shot, containing the police administrative area, an ID number, a timestamp plus longitude and longitude figures to three decimal places.

The Washington Post published its story in early November of 2013. Online the story ran at about 3,500 words, featured highly produced video, photos, and three interactives. The journalists presented their coverage with two aims; first to explain what ShotSpotter was and how it shaped Washington police operations, and second to interpret the data and give readers the chance to explore it and find their own narratives. The explainer section walked readers through the evolution of the technology, from a Californian inventor’s passion in the early 1990s to its current state and presence in 65 American cities. The Post reported that D.C.’s system had cost about $5.5 million over the eight years since it was installed in 2005. It described the microphones, their placement, and the software that interprets the sound waves, while giving a couple of examples of criminal cases where ShotSpotter became central to police investigation. As an illustration of the system’s technical complexity, the Post’s designers built a tiny game allowing readers to test their ability to identify the difference between gunshots and fireworks, using examples from the ShotSpotter database. Figure 1: By Bonnie Berkowitz, Emily Chow, Dan Keating, James Smallwood, The Washington Post The Post’s interpretation of the data started with the spectacular headline figure—“39,000 Shooting Incidents in the District”—and continued with other statistics in the written story, infographics, and interactives. The reporters noted that SpotShotter more than double the number of gunfire

incidents than there were official gun-crime reports. They wrote that Washington’s detected gunfire had dropped off by 40 percent since 2009, and isolated a few hot spots with even bigger reductions. The newspaper also built an interactive, which played a slightly different role in rendering the data. Its visualization needed to give users the ability to explore the data to find personally relevant views, while grappling with the fact that the location data was only precise to three decimal places. In latitude and longitude numbers, three decimal places is only enough to denote a location within a 100-meter area.1 The police had truncated the data to protect the privacy of Washington residents in the ShotSpotter coverage areas. The graphic designers’ solution was to produce a kind of gridded heat map. It represented the data’s geography, showed the intensity of the gunfire in various areas, but avoided depicting it with more precision that the data warranted. Figure 2: The Washington Post’s map, zoomed out.

Figure 3: The Washington Post’s map, zoomed in. Even though this was a data-backed story, in line with the Post’s normal protocol, Fallis supported the choice not to publish the underlying files, despite attracting some criticism. “It’s meaningless to most people; our job is to make sense out of stuff. It’s not like a nursing home inspection database where somebody can type in grandma’s nursing home and the reports are up there. We’ve done that kind of stuff. This one, frankly, I didn’t think was a very sexy click-magnet apart from the context we put it in.” Carlton Purvis, a D.C.-based journalist with an acute interest in security and surveillance, had a different view. He filed his own FOIA for the data, and posted it, along with a simple interactive map, on the website When users zoomed in on the PublicIntelligence map, they saw the data as dots plotted in a grid. PublicIntelligence’s overly precise mapping of data evoked some mistrust. One commenter wrote, “This graph is bullshit. Zoom in, all the data is evenly spaced. Complete bullshit. Next.”, which attracted the response: “The points are evenly spaced because LAT/LONG data is truncated to three decimals (for obvious reasons). This results in a  100yd. grid. Please apply a tiny amount of intellect before calling bullshit. Next.”

Figure 4: Carlton Purvis’ interactive on Public Both before and after The Washington Post’s story, newspapers from Atlantic City, N.J.; Oakland, Calif.; and many other cities have published stories about crimes the ShotSpotter system detects or how their cities fund installations. Although ShotSpotter keeps its full client list confidential, the company says it has more than 80 installations in place, of which more than 70 are in the United States. It lists two Brazilian cities as clients, and boasts of installations in four countries. Still, the story that Fallis and his colleagues published in The Washington Post appears to be the only ShotSpotter reporting that had access to the system’s data. Distinctive and Notable

  • Sensor data was used in a way unanticipated by its developers. The Washington Post’s story used the ShotSpotter data in ways that neither the system’s designers nor the police agency intended. As such, there were delays in exporting the data since the police did not use the data in aggregate; reporters needed the vendor’s help in producing it. Ultimately, what may have looked like deliberate data obstruction on the police’s part was simply the result of missing skills.
  • The historical alterations of the system limited analysis. Similar to the point above, The Washington Post had to contend with the fact that it simply couldn’t make broad comparisons over all the years that the ShotSpotter system had been in place. Again, the police used and deployed the sensors as a day-to-day operational tool, not one of analysis. When the department thought a certain neighborhood needed more or less coverage, it changed the number of microphones there. When new contracts were signed, the system was reworked.
  • Publically funded data wasn’t necessarily easier to obtain. Installations of ShotSpotter equipment throughout the United States (and in some other countries) used for police surveillance are funded out of the public purse. However, this didn’t necessarily mean the Post could get the data as a public records request. For good practice and to further inform this report, the Tow Center made public records requests from two other cities that have installed ShotSpotter. Atlantic City, N.J. police refused the request on the basis that releasing the data “could lead to the system being compromised as the location of [its] sensors is confidential. The exact coverage area is also confidential.” They told us, “The data you are requesting could be used to track the location of the sensors.” Camden, N.J. police initially said that they needed to get the permission of the ShotSpotter company itself. The Tow Center obtained permission, and, for the 13 subsequent weeks at time of writing, has been corresponding with Camden police representatives on a weekly basis to no avail.
  • Decisions were made about how much of the data to publish. The lead journalist on the investigation, David Fallis, opted not to publish the raw data. In contrast to some data journalists’ “show your work” practice, he saw risks in being held responsible for the accuracy and freshness of the underlying data, reasoning that the originating police department was a better source. He did help curious people contact police and form their records requests.

Lessons for the Industry

  • Make sure data visualizations relay your findings with clarity, even when the data has limitations. Although The Washington Post and Public Intelligence both mapped the same data, there was a notable contrast between their treatments. The Washington Post deployed its considerable graphics resources to produce a highly specific visualization approach, combining a heat map with a customized grid (that avoided overstating the data’s precision) and mouse-over information panels. Public Intelligence, on the other hand, mapped the data in Google Map’s default format— point markers. While this represented the data reasonably well when zoomed out, it was overly precise when zoomed in. Comments on its site evidence audience confusion.
  • The sensors themselves were newsworthy. In this case, the investigative and graphics journalists put a huge amount of effort into explaining the system itself. Those implicit news values suggest that sensor systems like ShotSpotter are, for a general audience, novel, opaque, and not generally understood. While journalists may find it worthwhile to use data from sensors to tell stories, they may also find readers are interested in new types of sensor systems, how they’re being incorporated into civic life, and their broader ethical, legal, and social effects.

Drones for Journalism: An Overview of the U.S. Industry

The field of drone journalism in the United States is unlikely to remain static. Just a year ago, there were very few known examples of journalists using drones to report local stories and we knew of only a handful of cases internationally. Recently, though, we’ve seen a number of journalistic works incorporating quality video shot inside the United States using high-end drones. We have more evidence of enterprising photojournalists experimenting with cheaper camera drones to cover breaking news and color stories. Beyond these uses by news industry employees, we have also seen drone footage by hobbyists find its way into mainstream media. The Federal Aviation Administration, which has regulatory authority over U.S. airspace, asserts the right to regulate small drones, but its sole prosecution to date was dismissed by an administrative law judge.1 The FAA immediately appealed to the full national safety board. Regardless of the eventual outcome of the case, additional bodies of law and ethical principles will affect how journalists use drones. A separate section of this report explores those issues in detail and describes the regulatory outlook. While the difficult legal environment has slowed journalists’ use of drones, it hasn’t stopped it entirely.

A Typology of Drone Use Despite the pace of change in drone journalism, rough groupings have emerged: high-end specialists, enterprising generalists, and “the crowd.” It seems credible to predict that this typology will last. The case study about NPR’s Planet Money and its use of a cinematography drone, found earlier in this report, represents the high-end style of the field. To film a portion of a documentary, the team hired SkySightRC, a small agency that uses tens of thousands of dollars worth of kit and flies as a twoperson team: a camera operator working in tandem with a pilot. ABC News used footage by the same company for its coverage of the Colorado floods in September of 2013. The advertising and movie industries employ similar companies, both domestically and abroad. Overseas, the BBC and the Australian Broadcasting Corporation take the same high-end approach—expensive equipment operated by at least two highly trained people with lots of attention paid to the safety of the people in areas where they fly. Considering the contested legality of professional civic drone use, it may be surprising that at least three insurance companies have been prepared to write policies covering professional camera-drone flights inside the United States. To understand generalists’ use of drones, a useful source is a survey run by the National Press Photographers Association in March of 2014.2 The results revealed that many newsrooms across the country are experimenting with small drones for journalism. The precise number is very hard to confirm: 172 survey respondents to the NPPA survey (who did not have to state their names or employers) said their companies had actually used a drone for newsgathering.3 Mostly, these were small consumer-grade camera drones costing hundreds, not thousands, of dollars. The Stories The subject matter of drone-shot footage varies. Professional photojournalists and camera operators have shot the aftermath of natural disasters and traffic accidents for breaking news. As color pieces, they’ve filmed spectacular architecture, crowd events, picturesque weather, and landscapes. An even larger number of respondents, though, said they wanted to use drones but were prevented by legal restrictions, safety concerns, and the expense. Journalists who fell into this group said they would use drones in the same ways as their bolder peers, noting additional benefits. A number of respondents said they would employ drones to get around obstructions; either for reasons of personal safety, police restrictions, or difficult environments. A few vivid examples:

  • “ Recently we had authorities burn down a house filled with explosives. A drone would have helped us get images of action going on at the scene and the fire itself.”
  • “ We did a video about the fifth anniversary of the closure of a GM plant. It would have been nice to go on an adjoining property and do a fly by to see the sprawling facility at more than eye level. Instead, we just drove by the fence and shot video.”
  • “ Drone aircraft could have given us better perspective on a story we are covering right now: ice jams in a river and the threat of flooding. We cannot safely get close-ups of the problem with a news crew. A drone could fly, give us a better angle and safer.”
  • “ Pretty much any story that police calls a crime scene.”
  • “ Major news events where access is limited and roadblocks prevent us from getting close enough to get the shots needed.”

The Economics Respondents repeatedly cited the lower cost of drones in comparison to helicopters as a significant benefit. In the words of one respondent: “For my helo costs, I could probably have a dozen HD-capable drones in the air.” Another wrote, “The cost of using a helicopter or fixed wing aircraft is high, anywhere from $400 to $1,800 per hour. A drone, such as the DJI Phantom, costs less than $23 per hour, making aerial news photography cost effective for daily use.” Around two thirds of the survey respondents worked outside the TV industry, meaning they hadn’t traditionally been able to pay the helicopter costs necessary to obtain aerial footage. However, that part of TV’s competitive advantage may be in danger. One respondent wrote, “Our competitors have helicopters and are [using them to get footage from news events where access is limited] every day. Although they are paying an expense my company won’t cover.” The Risks and Concerns When it came to the risks and problems associated with drone use, the main points of interest concerned privacy and safety. With regard to privacy concerns, NPPA survey respondents’ attitudes ranged from believing that it

was enough for journalists to follow current laws and ethical principles, to observations that these new tools with different characteristics mean journalists will have to rethink how ethics are applied. For example, one respondent said, “The media has legally, ethically, and respectfully used wireless microphones, tiny cameras, and super-telephoto cameras without invasions of privacy for decades. It is not in the interest of legitimate news organizations to alienate our audience by invading, or being perceived to invade, privacy. We will conduct ourselves with the same level of professional integrity with drones.” Another said, “You have no reasonable expectation of privacy when out in the open or in a public place.” Some respondents, however, could see new risks to privacy. “I don’t trust all of us to use a drone properly. With a helicopter, people can see and hear it. But with a drone, they can be virtually invisible.” Two responses articulated risks to safety associated with drones operating in a competitive environment:

  • “ There is a safety aspect. Police helicopters fly between 400–600 feet, media fly between 700–900 feet. I don’t trust that person flying a drone not to interfere with a manned aircraft trying to impress the boss. Today, we waited to the last minute to cancel a live shot due to lightning. We could see the storm on radar, but the reporter didn’t want to disappoint the producer and kill the live shot. I finally did. Safety first. But that reporter would’ve done the live shot, lightning be damned. And given a drone, who knows?”
  • “ It’s actually a very real concern. More and more newsrooms are being staffed by younger and younger people who have the ’get it on the air now’ mentality and the historical checks and balances of a newsroom just simply aren’t there anymore. I think drone use in a newsroom should be strictly regulated, much like satellite truck operations. Only one or two people in the newsroom should be authorized to fly a drone and I think the drone and the operators need to be licensed. I’m not in favor of letting everyone fly or providing a drone to each journalist. That’s very irresponsible.”

Drone Video From “the Crowd” News organizations also access drone footage from people who don’t formally work in the journalism industry. After an apartment building exploded in Harlem, N.Y., on March 12, 2014, the New York Daily News acquired footage shot by a 45-year-old business systems consultant named Brian Wilson. Lifestyle website The SFist embedded YouTube-hosted aerial video of a massive pillow fight in the city’s Justin Herman Plaza on Valentine’s Day of 2014.5 It was shot by filmmaker and drone enthusiast Beto Lopez, who has also flown his DJI Phantom above concerts, and sporting and community events. Self-described storm-chaser Brian Emfinger shot footage from his DJI Phantom 2 drone, flying it after a tornado destroyed homes in Little Rock, Ark., on April 27, 2014.6 By the next day, the video had attracted more than 1,600,000 views on Emfinger’s YouTube channel and the Associated Press had paid for reuse rights. At time of writing, YouTube reported that the video was played more than 2,400,000 times, well above the range of 500– 1,000 the typical video on a newspaper website gets, according to the Tow

Center’s “Video Now” research report.7 Emfinger distributed his video commercially using NewsFlare, a company that brokers deals selling usergenerated video content to media organizations. The three cases above are clear-cut examples of news companies using drone videos shot by people who aren’t their employees. There are, of course, other more ambiguous cases in which media companies used drone footage with one eye on the regulatory environment. The Spokesman-Review published a story on January 1, 2014 using aerial video of a community lake-swimming event. Jesse Tinsley, a photographer employed by the Review, said he shot it on his day off using equipment he’d bought himself. Both the photographer and the newspaper’s editor told journalism commentary site Poynter they expected they would therefore avoid legal action from the FAA.8 Respondents to the NPPA survey reported similar contrivances. Of course, not all news video reaches audiences through newsrooms. Drone footage flourishes on Vimeo and YouTube, and is publicized via Facebook and Twitter. Both of the major manufacturers of consumer drones showcase their customers’ videos in galleries on their websites. Mostly, the videos document topics outside the realm of journalism, although weather events and extreme sports, including surfing, motocross, and mountain biking, are popular subjects. As drone journalism in the United States takes shape, it has emerged with three broad production styles: specialist, generalist, and crowd. Each, however, has its own caveats. Specialist, high-end usage is rare, and not often publicized; lawyers and managers from the commissioning companies advise caution. Use by generalist news photographers appears to be constrained by the ambiguous regulatory environment; few newsrooms are making serious investments in camera drone training and equipment. Meanwhile, cheap camera drones are rapidly democratizing aerial video, some of which benefits news consumers, but further muddies the operating environment for journalists who are banned from doing exactly what amateur pilots can do legally.


The Third Section: The Legal and Ethical Considerations for Sensors and Journalism

Legality, Ethics and Journalistic Sensing: An Introduction

Up to this point, we have described the current landscape, laid out a history, and examined some notable sensor-based journalism. But, we hope, the previous sections of this report will quickly show their age, as enterprising and imaginative journalists use sensors to report new stories in different ways. These following essays aim directly at that. Reporters and newsroom managers should quickly see how the underlying ethical principles of journalism apply whenever sensors are being used. Also, they’ll see which laws come into play. Many of the authors that follow identify looming legal and ethical questions that journalists and their advisers will no doubt have to navigate, if they aren’t doing so already. While journalists working in the space of sensor journalism will not have to build new ethical foundations, these reporting methods do have distinct characteristics, some of which may be unfamiliar. Therefore, reporters may have less experience identifying the potential problems and their solutions. In commissioning the following articles, we have examined the distinctive features of sensor-based journalism and picked out some common places where those features intersect with legal and ethical issues. As a result, we offer four main groupings. The first set of essays here examine privacy and surveillance issues. The second set concern the laws governing building and acquiring this technology. Next, we take a look at the trouble that might arise when journalists place sensors and drones out into

the field (and how they might avoid it). Our final set of contributions looks at truth and the data from sensors—issues of accuracy, interpretation, and representativeness. In the realm of privacy and surveillance, media attorney Nabiha Syed explores the bodies of law that will govern what journalists are allowed to record with sensors, which issues will surface when they overstep the mark, and how the courts might navigate through this terrain making new laws as they go. Kathleen Culver, a journalism ethicist, looks at how communities have traditionally expected privacy and why journalists may want to adopt practices from other professions as they renegotiate power relationships between newsrooms and audiences. Josh Stearns, a journalist, strategist, and organizer, connects the emergent practice of journalists using sensors to the raging debate around state and corporate surveillance. He notes that by involving communities in journalists’ sensing projects, we may have obligations to ensure their physical and social safety. While few journalists have had to think deeply about patents, technology licensing, and whether their equipment will interfere with the communication systems used by ambulances and other civil services, Matthew Schroyer and Mike Hord have already expertly done so. Schroyer, a journalist and technology educator, steps through the arguments about why journalists might want to give away the intellectual property they create when designing new reporting tools (or when there might be a viable business case for holding it tight). Hord, an engineer at DIY electronics retailer Sparkfun, takes a look at how the regulator of the electronics industry, the Federal Communications Commission, is straining to keep pace with the booming “makers” movement, and how journalists can help them out. The physicality of sensors and drones brings operational risks. Diana Cooper, an attorney specializing in robotics law, raises a fascinating scenario whereby journalists who allow others to build on their hardware and soft ware designs may inadvertently open themselves to liabilities they haven’t considered. Thankfully, she identifies a middle path between open and closed licenses that might promote innovation and collaboration while still protecting contributors to a vibrant technology development community. Journalism professor and drone pilot Matt Waite teams up with the National Press Photographers Association general counsel, Mickey Osterreicher, to survey which bodies of state, federal, and constitutional law will affect reporters who want to use drones in the coming years. Deirdre Sullivan is an in-house senior counsel who has long experience helping the journalists, managers, and technologists of The New York Times know when they’re taking too many legal risks.1 She imagines a future when the laws and operating procedures have been settled for journalists using drones and sensors, at which point human error remains and personal injury and negligence claims are sure to follow. She examines how they work, and how the courts calculate breaches and penalties, and how companies might mitigate their financial exposure. Three authors have analyzed how journalists using sensors can uphold their ethical obligation to tell an accurate story. Lucas Graves, an assistant professor in journalism who specializes in fact-checking, opens up the process of truth-making in data journalism and shows where uncertainty and error emerge. Beth Stauffer, an environmental sensor specialist placed at the Environmental Protection Agency’s Innovation Team, describes how experts design their data collection so their information can be used to draw accurate conclusions. Finally, Lela Prashad, a data scientist who works with remote satellite sensing, reminds journalists to make their work richer by adding nuance and humanity in the space between and beyond their data points.

In an early exchange of essay drafts, Josh Stearns wrote the following words, which seem like the perfect description of the current evolutionary state of laws and ethics in journalistic sensing, and a prescription for how to proceed. “It is worth remembering that just as newsrooms are learning about the power and potential of sensors, so too are our communities. We are in the early stages of sensor technology, what Julie Steele of O’Reilly Media calls the precursor, the ancestor, to what will change our lives.’” At a panel on the ethics of sensor journalism organized by the Tow Center at Columbia University in the summer of 2013, Professor Joanne Gabrynowicz warned that poor sensor journalism at this early stage of the game could irreparably damage trust and create more fear of sensors amongst the public. It could also lead to bad policy that restricts the long-term use of sensors in reporting. Matt Waite likes to remind journalists, “Don’t do anything stupid. Bad actors make bad policy.’ Indeed, culture change often precedes policy change, so now is the moment to get it right when it comes to the ethics of sensor journalism.” “These questions are complicated by ongoing legal uncertainty at the intersection of media and technology, from the Federal Communications Commission’s debates about network neutrality to the Federal Aviation Administration’s review of its policies for small drones. At the Tow Center event Kord Davis, the author of The Ethics Of Big Data, reminded participants, ’It is always possible to act in accordance to your values but it’s never possible to act in accordance with the law that doesn’t exist yet.’ The law focuses on what you can and can’t do, but at a moment of legal flux it is more important than ever that we focus on what we should and shouldn’t do. That is the realm of ethics and as a practice that is rooted in the public interest and the common good; the litmus test for these ethical questions should be our communities.”

By Nabiha Syed When Samuel Warren and Louis Brandeis first wrung their hands over “Kodakers lying in wait,” they likely didn’t realize that the camera was just the first of many sensors that would come to document the human experience. Sensors allow journalists, as Matthew Schroyer explains, to “arrive at evidence-based conclusions in the course of a journalism investigation.” Both the “evidence-based conclusions” drawn and the process of “investigation” yield separate legal issues. This piece will address potential issues related to consent, which is an issue at the investigation stage, as well as defamation and false light, both of which relate to the publication of sensorsourced content. Defining Sensors At its most simple, a sensor is a device that detects and responds to some type of input—light, heat, motion, moisture, electronic signals, pressure— from the physical environment. Under this definition, cameras are sensors: A camera detects and captures light input to capture an image. But while courts have grappled with the sensor-as-camera for over a century now, especially in the newsgathering context, other types of sensors have not received the same attention.

It is an open question whether courts will approach sensors as a logical outgrowth of cameras, or will treat them as a new category entirely. On one hand, sensors can capture what happens in public, just as cameras can; some sensors are just capable of responding to input other than light. On the other, however, sensors pick up information that humans cannot discern— which is unlike the visible image, and the relationship between the camera and the public. Courts could very well construe this difference to mean that some sensor use is more invasive than cameras for image-capturing. Cameras are much better understood than sensors by the general public, leading to some issues with whether sensor use would violate some reasonable sense of privacy. Newsgathering with Sensors: Consent and Notification Issues American courts have long rejected liability for observation that takes place in public thoroughfares or in plain view, where one could be observed by passersby.2 This logic may be a starting point for courts’ reasoning on sensor use: So long as the sensors are in public, picking up information readily available in the public thoroughfare, the usage is legal. By this logic, explicit consent for sensor use is unnecessary, because the input detected by the sensor is public. The recent “right to record” cases underscore the value in recording what transpires in public. Several courts have now held that the First Amendment protects the right to record audio and video in public places, especially when recording the conduct of government officials.3 Importantly, however, the right to record does not mean that consent is no longer necessary for recording. A closer look indicates that existing caselaw focuses on public officials, and even then draws the line where there is a “compelling interest,” such as in border security.4 What the right does high-

light, however, is that courts recognize a fundamental interest in observing the public sphere, and are willing to extend protection accordingly. This interest may be helpful when thinking about using sensors to record particular types of input. Despite this trend, there are still two reasons that gaining consent or giving advance notification of the use of sensors is useful. The first is that courts could conclude that the unique capabilities of various sensors distinguish them entirely from cameras. After all, a camera senses and captures light to recreate, in fixed form, images that a person already has the capacity to see. That is, the sensor duplicates what a passerby can witness. This is not exactly true of many sensors, such as infrared or pressure tools. Courts may then refuse to extend the right to record to sensors as a blanket rule. The second is that any right to record or gather data—image or otherwise— is still subject to the longstanding principle that journalists are bound by laws of general applicability. As the Supreme Court has explained in Cohen v. Cowles Media Company, 501 U.S. 663 (1991), First Amendment rights are not offended when “generally applicable” laws have “incidental effects” on the ability of the press to gather the news. In short, the press has no special exemption from the law. That means that both traditional tort law and other statutory protections apply to the newsgathering context. The following are two arenas of particular note to sensor journalism. Trespass Asking permission before one attaches or otherwise uses a sensor may ward off potential trespass claims, which could come about in two ways. Most obviously, one could affix a sensor to property owned by someone else: This scenario is precisely what the Supreme Court considered in United States v. Jones, where law enforcement attached a GPS sensing device to a suspect’s automobile and used the device to monitor the suspect’s movements. At

least some justices found that attaching the sensing device did, in fact, run afoul of common-law trespass.5 This signals one simple limiting factor for sensor journalism: Don’t attach your sensors to property you don’t own. Similarly, a sensor-equipped device could also fly or flow through private property, picking up a possible trespass claim along the way. A drone or a bicycle, as was used by the “Beijing Air Tracks” project to detect urban particulate and carbon monoxide levels, might have exactly this problem if operating on or through private property. Of course, permission to be on private property may be implied by custom or by the nature of the premises, such as a restaurant open during business hours. Implied consent may also be granted when one enters private premises and is not told to leave, or when prior contact indicates that one does not need express permission. This is subject to a reasonable person standard, and even this implied permission can always be revoked. Importantly, trespass protects only against physical invasion of property.6 Should a sensor travel through public property7 and collect data originating from private property, one would likely not face a trespass claim. And even if trespass-by-sensor occurs, a sensor journalist would only be responsible for the physical harm done—injury to reputation or emotional distress from the resulting publication often does not count. It’s hard to imagine what palpable damage to property the sensor itself can do. What is worth more attention is the vehicle, if any, to which a sensor is attached—a bicycle or a drone can cause more tangible damage.

Wiretapping Wiretapping conjures images of intercepting audio8 telephone calls, not devices that detect dust or environmental pollution. But as the recent Google Street View Wi-Fi-sniffing case indicates, sensory collection that involves electronically transmitted data may run afoul of the Federal Wiretap Act. In the Google Street View case, the Ninth Circuit Court of Appeals determined that Google’s allegedly inadvertent collection of data—including usernames, passwords, emails, and documents—from an open (that is, non-password-protected) Wi-Fi network constituted a violation of the Wiretap Act. An exception in the Wiretap Act permits accessing electronic communications that are configured so they are “readily accessible to the general public,” which Google argued applied to unsecured Wi-Fi networks. Google also likened Wi-Fi signals to radio communication. The Ninth Circuit rejected these arguments, looking to what members of the general public might access: “[M]embers of the general public do not typically mistakenly intercept, store, and decode data transmitted by other devices on the network. Consequently, we conclude that Wi-Fi communications are sufficiently inaccessible that they do not constitute an electronic communication readily accessible to the general public.”

This logic could foreseeably extend to other types of sensory collection that gather electronic communications. Should a sensor be capable of such collection, it may be worth seeking consent of any possibly affected parties where the collection occurs. Printing Sensor-derived Content: Potential Defamation and False Light Issues At first glance, sensor-based reporting does not implicate reputational issues: If a sensor detects cicadas in a neighborhood, or measures the air pressure in a certain county, whose reputation is affected? And if the data is gathered in the aggregate, how would any conclusion attach to any single individual? Observations of natural phenomena that don’t implicate individuals or specific entities seem to skirt personality-oriented liability, and preclude claims like defamation and false light. The stories explaining sensory data might, however, be susceptible to these legal claims. Take the example of the drone flying over the Trinity River in Texas, whose camera picked up footage of a river full of pig blood downstream from a slaughterhouse. Imagine that—for follow-up purposes—you employ sensors in the water to detect the presence of animal blood or other contaminants. Your sensor detects particles of blood, and you conclude that based on past history, it is from the slaughterhouse. The slaughterhouse responds, arguing that there is another explanation for the conclusion you have drawn. You could easily face a defamation claim for communicating a false statement of fact that harms the slaughterhouse’s reputation. Defamation and false light claims are also important to understand because they will likely serve as proxy claims in jurisdictions, like New York, where common-law privacy claims are unavailable.

Defamation Truth is an absolute defense to any defamation claim, and a plaintiff carries the burden of demonstrating that a report is false. In any reporting that relies on sensor data, such as detecting blood particles in water, that data should be corroborated with further reporting to check the accuracy of any conclusions. One practice to avoid is any valorization of science: Sensorderived data should be verified and checked, just as a reporter would do with any tip from a source. As a general matter, a finding of defamation requires some degree of fault by the journalist. Under the First Amendment, the bare minimum standard of fault9 is negligence: A prospective private libel plaintiff would have to show that a reporter’s conduct was less careful than one would expect of a reasonable journalist in similar circumstances. Cutting-edge forms of newsgathering complicate this analysis, because courts do not have many “similar circumstances” or existing precedent to compare. Presumably, courts will want to examine both whether the source (that is, the sensor) was reliable, and whether the source’s information (that is, the data) was reasonably interpreted by the reporter. This implicates both the reliability and accuracy of data collection as well as whether it actually substantiates an allegedly defamatory conclusion. How would a court engage in this analysis? Courts might look to whether a reporter has experience using sensors or in data analysis, whether a sensor was properly calibrated, or whether there was any reason to doubt its accuracy. The outcomes of this analysis could be wildly variable, depending on the court’s resources. Developing well-defined industry standards of practice will be crucial in helping courts determine whether there was any fault in the reporting process. False Light Unlike defamation, a false light claim—asserting that publicity cast a plaintiff in a false light in the public eye—requires no defamatory statements, only that a plaintiff finds the statements to be offensive. False light protects emotional injury and personal feelings like embarrassment or helplessness, regardless of whether any tangible harm actually occurred. Some states, including California, find that even untrue implications can be the foundation of a false light claim, even if no explicitly false statement has been made (as is necessary for a successful defamation claim). Flawed data, or flawed explanations of robust and accurate data, could lead to false implications. Imagine, for example, using a GPS sensor to trace and identify bus routes in a given city, like what Digital Matatus is doing in Nairobi. This data might yield details that some bus drivers are driving erratically or inefficiently. Should the article later focus on a particular bus driver, or include a photograph of one, that driver could easily object by claiming that the article implied that he too drove in such a negative manner, which cast him in a false light. Proving false light is a high burden: The alleged statement must identify the plaintiff, be highly offensive to a reasonable person, published widely, and demonstrate some fault on behalf of the publisher.10 As discussed above, courts will likely face difficulty in parsing what fault looks like in the sensor context, precisely because there are few standards for comparison. To mitigate any risk, reporters should contemplate and closely analyze any implications that can be drawn from their work, including both the published data and its analysis.

Conclusions Sensors are not yet considered a routine newsgathering tool. This novelty may prove challenging to courts, which do not yet have the framework to assess the reliability or accuracy of sensors as a source, nor to evaluate whether a particular instance of sensor-based reporting was conducted reasonably. This is where ethics and industry practice must lead the way. Ethical guidance on how best to use, evaluate, and maintain sensors—and to interpret the resulting data—will not be found in the law. But ethical guidance and established industry practices can provide key instruction to courts, which will need assistance in determining whether a newsgatherer has relied on sensors in a responsible or reasonable way. In the meantime, the novelty and unique capabilities of sensors as a newsgathering tool may provoke legal claims.

Rights, Ownership, Ethics and Sensors in Journalism

By Kathleen Bartzen Culver When a reporter responded to news of a bomb scare at a Pennsylvania mall in January of 2012, he quickly turned to sensed data to help with sourcing. Using a social geotagging app called Banjo, Andy Stettler of The (Lansdale) Reporter searched for users who had checked in at locations in the mall and began tweeting at them for live updates from their locations. From them, he learned that only parts of the mall were being evacuated and that the threat appeared localized to one area. From Stettler’s process, we can quickly see the value of sensed data in even the most basic breaking-news reporting. But we can also highlight the ethical concerns that arise from data collected, even when gathered with consent. Banjo enabled Settler to contact sources and verify information, but it also would have allowed him to aggregate these sources’ data, tracking them across various stops in the mall, raising clear questions of newsworthiness and privacy. Citizens are subject to unprecedented gathering of their data—by government, by private institutions, and by news media themselves. This is enabled, in large part, by sensors embedded in physical devices—including digital activity monitors like Fitbit and the iPhone in your purse—connected in a vast network loosely called the Internet of Things. As the development of these devices outpaces laws, guidelines, and, at times, even understanding, journalists have an obligation to consider the ethics of collection, use,

storage, and protection of sensed data. They must ask questions about why we collect data, who controls this data, and what means will effectively empower individuals without subverting newsgathering. Data is increasingly easy to collect from a variety of sensors—video from a camera mounted on a drone,1 ground temperatures with a soil thermometer, and GPS location from a smartphone check-in.3 Journalists can employ their own sensors, as USA Today did when its investigative team used x-ray fluorescence devices to document lead contamination in soil surrounding old smelting factory locations.4 Or they can make use of sensed data from other entities, as the Florida Sun Sentinel did by using state toll pass data to uncover reckless driving by police officers.5 These kinds of reporting demonstrate the value of data in helping journalists serve the public interest and can illustrate useful parameters for ethical practice in collecting and using sensed data. Ethical Foundations of Privacy Use of sensors for data collection puts novel pressures on legal conceptions of privacy. We often premise privacy on the protection of personal space. We conceive of places where a person might have a reasonable expectation of privacy and punish intrusions into that space. By default, we afford less protection to any activity in public, presuming that people have no such reasonable expectation when they’re in a mall, restaurant, or town square. We also assume fewer protections for personal information that is surrendered willingly. But the ubiquity and sophistication of sensors and data streams have potential to transform these considerations. Because we are moving from simple human observation in public spaces to technological surveillance, practices raise questions with which courts and legislatures have not yet sufficiently wrestled. But news organizations have long used ethics to guide practice in the absence of, in addition to, and indeed in contravention of the law. As media ethicist Clifford Christians says, “Privacy is not a legal right only but a moral good.” Scholars conceive of privacy in a number of key ways, including the fostering of the individual versus the collective, protecting people from unwanted access, and enabling self-determination. Christians sees privacy as a central element to flourishing as individuals, ultimately critical to the common good. “A private domain gives people their own identity and unique selfconsciousness within the human species… Privacy as a moral good is nonnegotiable because controlling our life’s core is essential to our personhood.”7 Media uses of private information, then, must be justified in terms of this common good. This does not mean simply employing the shorthand of “the public’s right to know,” but instead deeply weighing losses of privacy and the effects on the common good. Key to Christians’ argument is the idea that the interests and evaluations that matter belong not to journalists but instead to the public. With privacy, he imagines a media ethics that is fundamentally concerned with the welfare of individuals.

Framing the Ethical Questions To date, media ethics has focused far more on the publication of private information. Sensors and data, however, raise concerns before, during, and after collection, regardless of whether the information is used. We must consider questions of individuals’ control over data about themselves. Daniel Solove’s work8 examining government data collection defines four key concepts that usefully frame concerns about sensing by journalists. Aggregation: Seemingly small bits of information can be fused into a much larger and clearer portrait of a person. For instance, MIT researchers recently reported the ability to uniquely identify 95 percent of individuals from just four spatio-temporal points in mobility datasets stripped of personally identifying information.9 Our patterns—a sort of mobility fingerprint— make us unique enough that our activities can be traced. Exclusion: Privacy is endangered when people 1) don’t know what information about them is obtained or how it is used, 2) don’t have the chance to opt out of collection, or 3) don’t have the opportunity to correct false or misleading data. Secondary use: Data obtained for a legitimate purpose can later be used for another reason that is not justifiable. For instance, public radio station WNYC asked listeners to send SMS messages reporting the progress of snow removal on their streets,10 netting not just the reports but also the participants’ mobile phone numbers. The listeners opted-in to offering reports, but their privacy would have been violated had the station later used the mobile numbers for fundraising messages.

Distortion: Data can paint a faulty picture of a person, often by reducing the whole to particular parts. Compensation and donor databases are often critiqued as distortions, reducing a person’s work life or political philosophies into mere dollar amounts. A Bill of Rights Approach Some argue that private organizations pose a greater threat to individuals’ privacy rights—and thus both their humanness and the common good— than government snooping and intrusion. We have Fourth Amendment rights and legal due process to address government violations. But what of Facebook? Google Glass? The Twitter API? “Privacy is rarely lost in one fell swoop. It is often eroded over time, little bits dissolving almost imperceptibly until we finally begin to notice how much is gone,” Solove writes.11 These concerns have led to calls for a consumer bill of rights for the Internet of Things. Connected devices, argues Adafruit Industries’ Limor Fried,12 should be open-source to empower users and deter abuse. Among her quoted core principles are:

  • Open is better than closed; this ensures portability between Internet of Things devices.
  • Consumers, not companies, own the data collected by Internet of Things devices.
  • Internet of Things devices that collect public data must share that data.
  • Users have the right to keep their data private.
  • Users can delete or back up data collected by Internet of Things devices.

But these ideas apply more neatly to devices, platforms, and apps than to reporting the news. Surely the speeding cops in south Florida did not have the right to keep their toll pass data—from publicly funded squad cars— private. Journalistic collection, use, storage, and protection of data are more aptly weighed by their justifiability using the following principles. 1. Narrowness: Rather than casting a sensing data dragnet, news organizations must be narrow in their focus. Certainly, sometimes stories will arise from gathered data. But more often, story ideas should dictate specific data to be collected and used. In the Sun Sentinel case, for instance, reporters learned of a high-speed crash involving a police officer and sought the toll pass data to investigate how widespread the problem was. They would be far less justified in a request for toll pass data for all Floridians, even if it was anonymized. 2. Minimization: Relatedly, news organizations should gather sensed data for these specific purposes and not later use collected material for other reasons without specific justification. 3. Transparency: People, not organizations, own the data points about themselves. If not, they have lost the core of their humanness. They must be made aware when data about them is being collected and stored. News organizations adhering to the common good cannot be opaque about their data collection. 4. Accountability: True media accountability—a responsible and trusting relationship with the citizens journalists purport to serve—is often dismissed with shallow references to newsworthiness or acting in the public interest. The public must play a meaningful role in defining acceptable datagathering and responding to its use.

5. Protection: Data collection and storage by journalists cannot be ethically justified unless it is protected from unethical uses. News organizations have an obligation to the common good to prevent intrusion by hackers, governments, and even seemingly benign secondary uses. Individuals have absolutely no control of their data if control has slipped from the fingers of a journalist. This protection is particularly critical when the data involves the exercise of such fundamental freedoms as speech, protest, political association, and religion. These principles sit on a foundation of ethics and a common morality, to be certain. But they also serve a self-interest for news organizations. Challenges to legitimacy and authority abound. Citizens use digital means to contest, and at times usurp, the roles journalism can play in a democratic society. We also see a public more concerned about surveillance by governmental and other organizations. During this time of debate, news organizations that open up their conversations about sensing and ethics are engaging in acts of self-preservation. In articulating and supporting such principles as narrowness and transparency, they have the opportunity to affirm their roles as standard-bearers. Emergent Ethics A common-good conception of privacy asks journalists to adhere to a common morality, rather than a narrow code of professional ethics. This results in clear tensions but can be seen to privilege certain professional traditions, including beneficence and minimizing harm. The protection, transparency, and accountability described above, for instance, appear to run headlong into journalistic notions of independence, avoiding conflicts of interest. Giving people control over their data, we could argue, subverts independence and entangles journalists with those they cover. But an ethics rooted in the common good would argue that harm to privacy can be minimized with this accountability, essentially trumping

the concern about independence. In short, it is more important for an individual to retain her ownership and control over the core of her humanness than for a journalist to be free of any conflict. Data practices in the age of the Internet of Things may require an ethics grounded in the very essence of these devices—interactivity. At its base, interactive means a device, program, or service that responds to user activity. We can see journalistic ethics similarly—a human exchange that respects the needs, will, and dignity of the people journalism serves. Privacy, then, is not a secondary concern, but rather a primary good defended for the welfare of all citizens.

The Ethics of Sensor Journalism: Community, Privacy, and Control

By Josh Stearns The rise of sensor journalism comes at a unique moment of flux in the debate around journalism ethics and raises its own set of questions and concerns. Three of the leading professional organizations serving reporters— the Society for Professional Journalists, the Poynter Institute, and the Online News Association—are undertaking major examinations of journalism ethics focused on how new technology and practices are changing the field. At the same time, we face complex and critical questions about society’s changing relationship with data and personal information. Revelations about both commercial tracking and government surveillance have sparked a contentious national debate about our fundamental rights regarding privacy and security. Meanwhile, community participation in the journalism process has blurred the lines between news professionals and the audience. It is into this public uncertainty and ambiguity that sensor journalism emerges. This social and political context is critical to thinking about how journalists work with, and collect data about, their communities. We should acknowledge that we are trying to define an ethical framework for a multifaceted and emergent practice. Sensor journalism, as currently constructed, already embodies a number of different forms and models, each raising different ethical questions. There are choices to be made both about what data to collect and what to do with that data once you have it.

Similarly, the ethical questions differ depending on whether you are working with willing or unwitting participants. We also need to account for the ethics surrounding sensor data collected by journalists, versus data gathered through crowd-sourced efforts for newsrooms. Below, you’ll find a focus not so much on answering questions but asking them—questions we can ask ourselves, questions we can discuss in our newsrooms, and questions meant to spark conversations between journalists and communities. Ethics of Community In announcing the Poynter Institute’s new book, The New Ethics of Journalism: Principles for the 21st Century, co-editor Tom Rosenstiel wrote, “Technology has so drastically altered how news is gathered, processed and understood, it has taken journalism closer to its essential purpose. […] News always belonged to the public. Now the ethics of journalism must consider the role of the audience, and the impact of community and the potential of technology more fully.”1 The justification for our nation’s strong protection of press freedom is rooted in the idea that gathering and disseminating the news contributes to the common good. Therefore, many ethical debates in journalism turn to questions of whether potential harms are outweighed by the public interest. So, as we consider the legal and ethical questions surrounding sensor journalism, the role of the public should loom large. However, since both our communities and our technology are in such a state of flux, this is no simple task. From the start, C.W. Anderson has argued, “The public was journalism’s totemic reference point, its reason for being.” Yet, in the digital age, while “the journalism-public relationship is still paramount,” the public is no longer a narrow, easily defined category— it is everywhere.2 Indeed, journalism is still produced for the public, but it is also increasingly produced with the public. Passive Participants and Active Transparency There are at least two communities to consider in the design of sensor journalism projects: The members of the community where we are collecting data, and any community members who are helping us collect that data. There is still a lot people don’t know or don’t understand about sensors, so we should approach our journalism with as much openness and transparency as possible. Trust takes a long time to build, and a very short time to break. Because of both the perceived and real concerns to privacy inherent in sensor journalism we should not rely on passive transparency, but instead must invest in a more active kind. Passive transparency means sharing the minimum amount of information possible, and requires communities to seek that information out. Instead, an active transparency approach calls on us to engage in proactive outreach and education about our sensor use. Whenever possible we should make the work of sensor journalism visible to those who might encounter our physical tools. We can do so by labeling sensors with signs to make them visible, by publishing maps of where the sensors are placed online and in print, by attending community meetings to explain our projects, and—if we are working in radio or TV—we can talk about the sensors on the air and point out where they are gathering data. Ideally, this public awareness work can and should begin before the sensors are deployed.

Kord Davis gave a concrete example of the sensitivities of transparency during the 2013 Tow Center sensor journalism summit. Davis pointed to the use of aerial drones to collect geocoded data in Africa that may include indigenous populations whose values—when it comes to the collection and tracking of that data—we don’t know or understand and who have not been part of the design of our study.3 While this may be an extreme case, Davis’ point is that our newsrooms and communities have values, and those values may not always be in alignment. As such, our ethical framework has to account for both our professional norms as well as the values of the communities with whom we are working. Data from the Crowd, Responsibility for the Newsroom Some of the most exciting early projects in sensor journalism, such as the WNYC “Cicada Tracker” project, have included crowdsourcing elements that engage communities deeply in the process of journalism and data collection. As more and more newsrooms look for meaningful ways to connect with their communities, sensors offer an exciting, hands-on opportunity for collaboration. However, employing the crowd to help collect data also raises new ethical questions about how to prepare and protect the community members with whom we work. While we need to communicate and discuss these issues with community participants, newsrooms also have to consider issues like the legal and safety risks participants might face in collecting data. An Online News Association working group looking at the ethics of usergenerated content and social journalism asked a series of questions that are also relevant here: “Is it okay to ask members of the public to compromise their safety in order to gather information or content for the news media? Or even to accept material from them when they choose to take such a risk?

What are our responsibilities to them?”4 For example, if we are working with communities on a project like Public Lab’s water quality testing effort, what if one of the volunteers drowns while trying to place his or her sensor, or is sued for trespassing on private property while sojourning to the river? If we treat community as an end, not simply a means, as Poynter suggests in its new book, we have to understand that our ethical responsibility extends beyond the subjects of our stories and the community of our readers to the collaborators with whom we work during crowd-sourcing efforts.5 This approach demands that newsrooms think about how they train, support, and engage their communities, all of which can actually help create more accurate journalism and more deeply connected audiences. The Ethics of Privacy and Control in an Age of Surveillance It would be impossible to discuss the intersection of sensor journalism and our communities without addressing the impact of recent revelations about privacy and surveillance by governments and corporations. While the privacy issues that sensors raise exist independent of factors like the Snowden revelations, news about the National Security Agency (NSA), and commercial data collection and tracking, these are still deeply influencing the national dialogue about privacy and security in the digital age. While public response to the NSA revelations was initially muted, as a fuller picture of the agency’s surveillance programs emerged, public polls shifted dramatically. The Snowden documents—paired with security breaches at major retailers like Target—have spiked interest in digital privacy and security. In many respects, sensors may seem even more intrusive to the general public than the NSA’s bulk collection of metadata. Sensors collect data about our movements and our environment, which for many people feels much more immediate than recording what we do on our phones and computers. Journalists are late to the big data game and we are entering it amidst increased questions and concern about surveillance in our lives in general. During the Tow Center’s 2013 sensor journalism ethics panel, Robert Lefkowitz, the CTO at Sharewave, argued that the scope of the ethical questions we face are directly related to how we define sensor journalism. If sensor journalism is only about collecting environmental data, “then the issues are fairly clear,” he said, but once we begin collecting personal data they expand exponentially. Of course, defining the lines of privacy and personal information is increasingly challenging. At the same event, another panelist, Professor Joanne Gabrynowicz, pointed out, “The whole idea of public and private spaces is being turned on its head.” This concept is reinforced in an important paper on data protection laws by Eloïse Gratton, in which she writes, “A literal interpretation of the definition of personal information is no longer workable.”6 Changes in technology and in the amount of data people leave behind, both intentionally and unintentionally, are challenging what we mean by personal data. As the power and accuracy of sensors expand and tools for parsing data sets develop, analysts and algorithms can connect the dots in ways that can turn seemingly anonymous information into personally indefinable data. In her paper, Gratton argues that, given these changes, a literal definition of personal data can create three contradictory consequences when used to guide ethics and policy. The literal definition of personally identifiable information could be too inclusive (threating the free flow of information), too lax (misunderstanding how the context of individual data points), or simply too vague (leaving uncertainty regarding the point at which a piece of data becomes personally identifiable).7 Thus, returning to Lefkowitz’s initial point above, it becomes more difficult to differentiate “benign” environmental data, for example, from “sensitive” personal information. Instead of drawing arbitrary lines based on outdated notions of personal information, Gratton suggests we focus on the purpose of the protection—to avoid harm. Using risk of harm as a guide in thinking about how we collect and handle data has its own layers of uncertainty and interpretation, but it resonates both with the community-focused approach outlined earlier and with journalism ethics that have long grappled with balancing the needs of privacy with the public interest. In a very real way, sensors are expanding the web of surveillance, and so we need to consider the various risks of harm embedded in our actions. Much as journalists should base their own security precautions on real-life threat models, we should assess what threats our sensor projects pose to our communities. This raises important questions for journalists regarding how they collect, store, and protect the data they gather (as well as legal questions around how governments can and will request access to that data). Controlling the Data In a 2012 article, Alistair Croll argues that big data is this generation’s key civil rights issue. He points out how personal data collected on and offline is being used to make discriminatory decisions in everything from credit cards to loan offers. “Data doesn’t invade people’s lives,” he writes, “lack of

control over how it’s used does.”8 Indeed, the question of who gets to control the data collected about us was also a key point in a set of recommendations delivered to the White House in early 2014 by a coalition of civil and human rights groups. “Individuals should have meaningful, flexible control over how a corporation gathers data from them, and how it uses and shares that data,” wrote the groups.9 Both Croll and the civil rights coalition recognize that data can be a powerful force for positive change in health, environment, and social justice, but that without real agency from communities it can also be a double-edged sword. “While data is essential for documenting persistent inequality and discrimination,” said Wade Henderson, President and CEO of The Leadership Conference on Civil and Human Rights, “no one—no matter their color, ethnicity, or gender—should be unfairly targeted by businesses or the government for dragnet surveillance, discriminatory decisions, or any other unwarranted intrusions.” It is easy to see how journalists bringing sensors into a community might be seen as an unwarranted, or at least unwelcomed, intrusion. We have to understand the history and current state of surveillance in diverse communities around the United States, and the discrimination and threats linked to that surveillance.10 Our decisions have to be informed by that history and in conversation with communities.

In addition to the transparency and community engagement outlined above, whenever possible journalists interested in employing sensors should develop meaningful ways for communities to access, control, and discuss their participation and data. Developing these systems will only come through hands-on application, and will demand flexibility based on the journalism project, the newsroom, and the community. A few questions for consideration as newsrooms develop sensor journalism projects include:

  • Who has control over the data we collect (both during collection and afterwards)? Can communities and individuals access the data?
  • Can individuals and communities opt-in or opt-out? When is such an opt-in process important and when does the public interest outweigh the need for an opt-in?
  • If publishing open data sets, can we (or our communities) control whether or not the data gets used for commercial purposes?

Protecting the Data A key part of how we control the data we collect relates to how we store and protect it. There has been a growing emphasis on transparency and calls for data journalists to publish their data openly on the Web. While that level of transparency for most data should be the standard, depending on the data collected by our sensors and the feedback from our communities, we’ll need to think about how and if we should share what we collect publicly. In the planning stages of a data journalism project, newsrooms should discuss their policies and procedures around anonymizing and sharing data. Part of that planning must also address how it will be stored. Newsrooms need to understand the best practices for safely storing information, and be up front about how long the data will be kept and for what purpose. In his article on big data and civil rights, Croll notes, “If I collect information on the music you listen to, you might assume I will use that data in order to suggest new songs […] But instead, I could use it to guess at your racial background. And then I could use that data to deny you a loan.” This example implores us to question how we’ll control future uses of the data we collect and store. Also, when our sensors are our sources, there will be times when we need to consider how we will protect that data just as we would protect a source. Jonathan Peters, the press freedom correspondent for the Columbia Journalism Review, has written about the security and legal risks of journalists’ increased reliance on cloud storage for files and data. “The legal protections for journalists using the cloud is, well, cloudy,” he writes. Historically, the Privacy Protection Act of 1980 protects journalists from government searches and seizures but Peters notes, “It’s unclear whether the things journalists store in the cloud enjoy the same legal protection as the things they store on personal computers, on local servers, and in desk drawers.”11 In the early days of sensor journalism it might be difficult to imagine a project whose data set could be of interest to government investigators, but in many ways, that is the point. We should grapple with these questions now and develop ethical guidelines for sensors that address all the kinds of data we collect—recognizing that, as our identities are increasingly defined by our data, more and more of that data is personal. Conclusion The intersection of changes in technology and shifts in newsroom processes and ethics mean it is nearly impossible to define any clear-cut guidelines for sensor journalism ethics. Instead, we can and should identify values and questions that can guide us appropriately. Recognizing that sensor journalism projects will vary greatly in project design, the technology they use, and the context of the communities where they are working, our ultimate goal should be to align those various elements in ways that build both new knowledge and safer communities. Alister Croll suggests that one of the most dangerous things we do when we collect data is “collect first and ask questions later.” In this model, he argues, “This means we collect information long before we decide what it’s for.” If we don’t ask these questions in advance there is no way we can provide our communities any control, or ensure that we are protecting their privacy and storing their data securely; Croll suggests “This act of deciding what to store and how to store it is called designing the schema, and in many ways, it’s the moment where someone decides what the data is about. It’s the instant of context.”12 While I’ve referenced the ongoing ethical debates and inquiries at the major journalism institutions, the questions outlined here need to be better integrated into those debates. As a field, we need to weave these threads together and consider ethical guidelines that are not so narrow that every new technology challenges them and sends us back to the drawing board. To that end, we should also draw on the expertise of those groups working at the intersection of press freedom and civil liberties like the American Civil Liberties Union, Free Press, and the Electronic Frontier Foundation. These groups, which have long worked with questions at the heart of journalism and individual privacy, can add a lot to our discussions and serve as a check against our industry’s assumptions and culture. If sensor journalism is to expand and thrive, however, we’ll need communities, elected officials, technologists, and newsroom stakeholders to buy in. It is not enough to define the ethics of sensor journalism through the lens of newsrooms and journalism ethics; we must also shape our choices through the lens of communities and their values. We need to be aware of historic issues around power and racism that are also embedded in debates about surveillance and move forward in a way that invites more people to the table to discuss these issues. This is an organizing challenge and we should rise to meet it.

Sensor Journalism Through Open and Closed-source Initiatives

By Matthew Schroyer Introduction Traditionally, journalists have relied on government agencies or selfreporting industries to obtain data for investigations. A new wave of rapidprototyping tools—including low-cost sensors, and easily programmable microcontrollers and microcomputers—has lowered barriers for creating data-collecting devices, potentially allowing journalists to investigate independently of those traditional institutions. Whether a particular sensor journalism initiative will succeed largely depends on how sensing devices are planned, designed, produced, and deployed in the field. Key to many of those decisions is how “closed” or “open” a sensor journalism initiative will be to collaborators and end users. While this section may be most applicable to journalists who are considering building their own sensing equipment, it is important to note that not all sensor-based reporting will require making your own tools. Stability, Rigidity in Closed-source Sensing Devices Although the share of open-source tools in newsrooms is increasing; from word processors and page design applications; to cameras; smartphones; and tablets the bulk of digital tools that journalists use on a daily basis are

closed-source.1 The success of the closed-source model is largely credited to IP-holders’ defined ownership, which can be wielded to dictate the future of the project or company.2 Ownership provides a clearly outlined revenue stream, which can be used to attract investors or secure talent who also wish to enjoy a stake in the project. End users also enjoy a distinct line of accountability from the developer, whom they can rely upon for service, patches, and upgrades. Closed-source devices may provide several important advantages for those looking to adopt, not develop, sensing tools. Namely, agencies might license a standardized, laboratory-tested sensing device that is managed through a polished, dedicated computer or mobile app. While most commercially available sensors, such as Geiger counters, air quality monitors, and pH meters, are closed-source and lack internet connectivity, the landscape is poised for change as more connected “Internet of things” (IoT) products come to market. The Nest Protect, for example, transmits carbon monoxide and smoke readings to a central server, which are accessible through a mobile app.3 Fitness trackers, like Fitbit Flex and the Nike FuelBand, sense motion, heartbeat, and GPS location, and can sync this data with smartphones or computers. Still, journalists may find closed-source sensing tools prohibitive in a number of ways. First and foremost, by its very nature, closed-source technology cannot legally or physically be modified to fit specific applications. Though there may be no stopping a curious individual from probing proprietary hardware, as is the case with “circuit bending” or “hardware hacking,”4 schematics are closely guarded and proprietary hardware is protected against unauthorized duplication. Additionally, data obtained from closed-source devices might be stored in a proprietary file type, which may require expensive software to access, process, read, or export it, making the data of little use to third parties and jeopardizing a news agency’s ability to be algorithmically accountable.5 (Fergus Pitt, personal communication) Furthermore, existing commercial technology may not be optimized for journalistic purposes. The Nest Protect, for example, is designed only for remote home monitoring, and fitness trackers may only be useful within the scope of a dedicated fitness-tracking application. For journalists seeking to make their own customized sensing solutions, going “closed” presents challenges with regard to development and servicing. Advances in rapid prototyping have reduced hardware development costs,6 but commercial software and hardware markets often demand high upfront costs to develop a product. Additionally, end users will likely expect you to provide troubleshooting services, product updates, and server space for cloud data storage. Developers also run the risk of becoming involved in patent lawsuits, which have become more expensive to litigate over time.7 While these costs may be distributed through the market, end users may face prohibitive license fees that could ultimately limit the size of the product’s potential user base. For journalism-friendly technology to exist in the commercial market, there must be demand; as of yet, it is uncertain whether media companies will find a market to support in-house development of commercial sensing technology.

Open-source Tools for Sensor Journalism Open-source dates back to (at least) the 1950s and 1960s, when user groups began to distribute and exchange software source code.8 Current database journalists might have familiarity with open-source software like the Open Office Suite, geographic information system (GIS) tools such as QGIS,9 and the ubiquitous Python and Ruby libraries. However, it’s much less common for them to be using open-source sensing devices in the newsroom. That’s not to say that a number of open-source sensing devices don’t exist that can be purchased as a kit, and assembled and configured with the help of digitally distributed instructions. A few can even be purchased as finished products that can be deployed with minimal setup. The Air Quality Egg (, for example, is a commercially available, open-source device, which registers carbon monoxide and nitrogen dioxide readings and sends this data to a cloud storage and graphing service. The Smart Citizen Kit by Acrobotic Industries ( smart-citizen) is another device on the market that measures air pollution, in addition to light and sound levels. Safecast (, a volunteer radiation-monitoring organization founded after the 2011 Fukushima Diiachi nuclear disaster, hosts crowdsourced, geotagged radiation readings and provides information about how to build and code Arduino-based Geiger counters. The DustDuino (, which is being tested as a sensor journalism platform, produces EPA and EU-compatible readings for coarse and fine particulate matter. Each of these low-cost sensors used in open-source projects has been criticized for its inadequate sensitivity levels, low resolution, and lack of precision and accuracy,10 however, low-cost sensor technology is improving. Some sensors coming to market (such as optical particulate sensors) have been compared favorably to more expensive ones.11 For those who plan to launch their own open-source sensor initiatives and create new devices, the community of end users will take on additional dimensions of importance when compared to its role in closed initiatives. The success of closed-source sensor projects is limited by product cost and utility to end users, whereas the potential of open-source projects is defined by the community that forms around it. Buy-in to this community is limited by the prerequisite hardware and software development skills necessary to build and contribute to the project, the time and energy necessary to learn the prerequisite skills, and the kinds of returns the end user can expect.12 As the user/developer group assumes the role of refining, releasing, and servicing the software or device—sometimes in novel ways—there is no defined end-of-life baked into the design of the software or device. While product development or deployment can be supported by multiple, simultaneous funding models, such as ancillary services, consultancy, freemium, or crowdfunding a physical product,13 the development community will always dictate the future of the device. Thus, the critical question for the project becomes: Are there enough sensor journalists, or other interested parties (i.e., citizens, hacktivists, and researchers),willing to shoulder the burden of development (in terms of time and material resources) to keep the barrier to entry low enough so that others may buy-in? A defining feature of open- and closed-source sensor journalism initiatives is a level of public education and community engagement. In sensing projects such as Safecast, the Air Quality Egg, WNYC’s “Cicada Tracker,” and the DustDuino, education about the environment and sensing technology was necessary to achieve community buy-in and recruit developers.14 Indeed, research has shown that those who work on open projects receive some kind of learning benefit in the process.15 A skilled, strong community of developer-users not only benefits the efficacy of the end product, but also provides transparency and lends the project algorithmic accountability. Where sensors may be used in critical journalism investigations, the public will scrutinize the accuracy and precision of sensing devices, making accountability necessary. Of course, open-source initiatives aren’t without drawbacks. Perhaps the biggest threat to open-source projects is the unpredictable nature of development communities.16 There is risk with investing in development and taking any product to commercial market, and crowdfunding can be unpredictable and difficult to master.17 The absence of copyright and claims on intellectual property rights limits access to more direct, reliable revenue streams, and opens the door for imitators to push out copies at marginal cost.18 Finally, giving a project’s destiny away to the user community also means granting others the power to influence the final product, which could also be an advantage if the community generates a novel and useful product.

Ultimately, no sensor journalism initiative has to be entirely open- or closed-source. The creators of the Arduino, for example, have published schematics and code, but retain the Arduino trademark. Likewise, where precision is paramount, closed-sourced sensing devices could be integrated into a project where the resulting data is open-source. However, collaborators may find they have no choice but to keep projects open. In his article, “How Open is Open Enough?” Joel West identifies four key motivations that drive developers to open-source their projects:19 1) Market share is lower than the minimum efficient scale necessary to support proprietary research and development; 2) There is not enough market power to resist buyer demands for open standards; 3) “ Tipping” of the standards contest is in favor of the open standard, making it infeasible to establish (or maintain) a proprietary standard; 4) A decision is made to accept commoditization of the particular architectural layer and shift competitive advantage to another architectural layer. More important than prescribing broad rules for what kind of projects should be open and which should be closed is recommending that an organization create a strategy to address patentable technologies where they may arise, while working to create a vibrant community around environmental sensing. Journalists may find that their own values, in terms of facilitating free flow of information and civic participation, align better with the open-source ethos.

Makers, Journalists and the Federal Communications Commission: Why the Laws Exist

By Mike Hord For journalists who are considering building their own sensors the laws governing electronic devices may seem restrictive and somewhat counterintuitive. However, solid principles underpin the rules, even if the fast pace of change in modern hardware development appears to be stressing the current regulatory regime. One category of sensor-based journalism projects runs the greatest risk of breaking these laws: projects that incorporate custom-built electronics, especially if they’re being deployed en masse. The Federal Communications Commission (FCC) has stewardship over the electromagnetic spectrum within the United States. The FCC regulates all devices that emit any kind of radio signal either intentionally or unintentionally, and issues licenses to people or commercial entities that market or operate those devices. In some cases, where the stakes are relatively low, laws simply require that devices or services using radio frequencies not interfere with higher priority systems. To give a common and probably familiar example, a device that broadcasts music from an iPod to a car’s FM radio can acceptably interfere with radio reception up to a couple of car lengths away; this is not generally considered an issue. However, if that

device were to interfere with an emergency band radio used by a fire department, this would constitute a serious problem and the manufacturer (if not the user) would likely be subject to significant criminal liability. The FCC also allocates the use of radio spectrum to commercial entities, such as television and radio stations, and cellular phone service providers. The commission takes interference with licensed uses of the spectrum, unintentionally or intentionally, very seriously and may even investigate and bring criminal charges. Fortunately, the design of most commercially available devices makes it very hard to accidentally interfere with licensed operations. The FCC operates under rules that have not been revised since 1989. They were written long before the DIY electronics revolution of the last ten years. But every day, companies like my employer, SparkFun Electronics, make it easier for individuals to produce small quantities of increasingly complex devices. The complexity of these devices routinely rivals that of the consumer products for which the FCC rules were originally composed. This new state of affairs has produced some ambiguities in interpreting current FCC wording. For instance, authorization from the FCC is required prior to marketing or importing a device for sale. Meanwhile, a small number of devices may be made for personal use. But we can easily imagine journalists’ work falling somewhere in between: What about a newsroom making a small number of custom sensors and distributing them for free? Or a for-profit consulting business that produces such sensors for a reporting project? So, why this lack of movement? The rules under which the FCC operates are laid out by an act of Congress; thus, an act of Congress would be required to change them. There is some leeway in how the rules are interpreted, however, and the FCC periodically releases white papers detailing its interpretations of key points within certain rules.1 More information on interpretation can be found by examining some of its past enforcement actions.2 But, we can expect that creative technologists will continue to come up with new scenarios that aren’t even covered by the guidance papers, let alone legislation from 1989. This section will help readers identify the gray areas, and navigate a path through. Types of Devices Devices can be categorized as one of two types: intentional radiators or unintentional radiators.3 Unintentional radiators are devices that do not specifically use radio waves to transmit any information. These are items like televisions, DVD players, and radio receivers. In general, these devices must be tested before going to market to ensure that they will not interfere with other devices. (There are limits to the amount of radio waves they are allowed to emit during use.) Although there is a legal requirement that these devices be tested, they do not have to be registered or reported to the FCC before they can be marketed.4 Intentional radiators are devices that use radio waves for information transfer. This includes those like remote controlled toys, cellular telephones, Bluetooth headsets, and any device using a wireless Internet connection. Intentional radiators can be further subdivided into the band, or primary frequency range within which they operate. Some bands require that those using devices in a specific range be licensed or trained. For example, maritime radio bands are restricted for use by persons communicating to or from ships and boats. Other bands, particularly the so-called industrial, scientific, and medical (ISM) ones, are intentionally allowed to be “dirty”— meaning devices operating in them, which include essentially all consumeruse goods, are expected to comply with certain rules about signal strength but may interfere with one another. Intentional radiators must be tested and registered with the FCC prior to marketing.5 Those marketed in the United States must also have an FCCissued ID code to be legal; failure to obtain or display this ID is increasingly common among inexpensive goods imported from overseas. Restrictions on Homemade Devices The FCC offers an exception to its rules about device testing to allow for private construction of up to five devices not intended for market sale.6 Obviously, this does not exempt the private citizen from complying with other rules; just because a cellular phone jammer is homemade doesn’t make it legal, and failure to exercise due diligence in the design and implementation of a device that causes unacceptable interference can result in legal penalties. Personal Meets Professional There exists a swelling space between commercial and private gadgetry, wherein a person could, on contract, create devices for a commercial entity to use. It’s practically possible, for instance, to create a network of distributed sensors reporting data over a cellular phone network for several hundred dollars or less. Clearly, devices created under such circumstances would not be marketed, as they are intended for the use of a single client. However, they can’t truly be defined as for personal use either. The FCC’s rules do not currently seem to exert guidance in this area. The current laws date back to a time when

transistor radio kits were state of the art;7 when devices of this complexity certainly could not be readily created outside of commercial enterprises. Until the FCC updates its rules, releases some guidance or interpretation on the matter, or investigates some relevant cases, we can assume that such non-marketed devices are legal, so long as they are constructed and operated with due diligence and good engineering practice to avoid any interference with other users of the spectrum. Another consideration falls within the realm of distributing electronic devices to the audience of a commercial entity. This is more clear-cut—such devices must be considered as “on the market” and dealt with appropriately just as any other commercial device. If the device is provided in kit form, however, the FCC has issued guidance to suggest that it need not be subjected to the same sort of scrutiny as a finished product,8 since the end user/ assembler of the kit is expected to have some expertise in the matter. Kitting is only loosely defined by the FCC;9 the agency has, however, specifically indicated that the mere act of installing circuit boards into an enclosure is not technically rigorous enough to grant exemption as a kit.10 Consider, for example, a distributed sensing project where a media organization invites readers to use a sensor and contribute data to a larger project. If the sensor package is fully assembled and ready to use, it would be subject to the same rules as any other commercial product. If it were provided as a kit, which requires some skill (even basic soldering) to assemble, that requirement would not apply.

Dissemination of Information Thus far, the freedom to distribute designs or data that can be used to construct illegal devices has not been restricted by the FCC. Good examples of this are cellular phone jammers; designs for such devices exist openly on the Internet and can be downloaded and replicated easily.11 Possession or use of such devices is, of course, still treated as a crime. Other information is even less regulated. Books and magazines, such as Forrest Mims’ excellent series Engineer’s Notebooks and the magazine Nuts and Volts, have been providing information about circuit design and construction for decades. There is an entire industry, dubbed “open source hardware,” which exists solely to provide instruction and supplies to those wishing to dabble in electronic design. Journalists, therefore, should consider dissemination of design information and details to be completely acceptable, and indeed laudable. Ultimately, it is the responsibility of anyone attempting to duplicate the device to ensure that they do not violate the FCC rules. Guidelines for Ethical and Legal Conduct Given that the rules do not explicitly forbid the creation of sensors or instruments as tools of journalism, we must consider what sort of guidelines we might follow to ethically and legally employ these devices.

  • First, do no harm. Under no circumstances should a device that intentionally interferes with legitimate uses of the radio spectrum be employed. This should rule out purpose-built devices, such as cellular or GPS signal jammers; misuse of standard devices (radio transmitters tuned to a frequency and left in transmit mode indefinitely); or abusive modification of legal devices (small FM or AM transmitters with signal boosters).
  • Seek knowledgeable advice. More than enough information exists in the marketplace to allow even relatively unskilled people to create devices that might fall afoul of various rules. When using electronics that actively transmit, seek advice from someone with reasonable expertise on the matter.
  • Avoid illegally marketed electronics.12 There are a surprising number of them in the marketplace these days, as unscrupulous foreign manufacturers sell to uninformed domestic suppliers. Of particular concern are remote video surveillance systems; the FCC has already investigated (and fined) importers and users of these devices. Remember, all devices that transmit wirelessly must display an FCC-ID number in order to be marketed legally in the United States.
  • Use FCC-certified radio modules.13 The FCC offers manufacturers the option to certify radio modules. This is meant to make it easier for a system assembler (say, a laptop computer maker) to integrate wireless functionality. These modules are pretested to ensure that they won’t interfere with legitimate spectrum users. By implementing them in a project, the odds of causing interference reduce dramatically. Modules for Bluetooth, WiFi, and cellular networks can all be purchased.
  • If challenged, stop using the devices immediately. In the unlikely event that an FCC enforcement agent questions your device, stop using it right away. The current state of affairs is such that an FCC enforcement agent is likely to become involved only in cases of serious infraction, where interference with licensed use of the radio spectrum is repeated and extremely noticeable.

The FCC Does Not Have Worldwide Jurisdiction It should be noted that this section pertains only to activity within the United States; other countries have different rules and laws. While Canada and the EU have relatively similar laws to those in the United States, Japan has somewhat more stringent guidelines. It is advisable to seek local advice before attempting to use any custom fabricated or non-commercial electronics in other countries.

The Traps of Open Source for Sensor Journalism

By Diana Marina Cooper As drones and similar technologies equipped with sensors become more available, journalists and media companies will have many legal and ethical issues to consider—including deciding whether the technologies they procure, develop, and use ought to be open or closed. At first glance, the journalism industry’s commitment to transparency and dissemination of knowledge suggests a natural allegiance with the open movement. However, the increased liability risks associated with open technologies could certainly sway the industry toward closed assets.1 This section outlines legal and ethical issues associated with open sensor technologies, and proposes industry-driven solutions aimed at mitigating liability risks and promoting ethical use. It proposes the development of an ethical licensing model to allocate liability and reduce the risk of harmful and unethical applications.2 As a complementary measure, this section suggests the adoption of an ethical code of conduct to govern the procurement, development, and use of open-sensor technologies by members of the media industry.

Links Between Journalism and the Open Movement Historically, the media industry and the open movement have exhibited a shared commitment to the dissemination of knowledge; both subscribe to the ethical ideal that information should be freely shared. Emerging trends within media, including the rise of participatory journalism, have nudged the industry even further along the ideological spectrum toward openness. Traditionally journalistic work was polished and “presented to the reader in its complete, hopefully perfect form.”3 In contrast, the current shift toward open journalism reflects the notion that a journalistic work is a continuing process that “encourages reader participation from the start, and even after the story is ’finished.”’4 This characterization of traditional and open journalism parallels the distinction between closed and open designs. While proprietary developers distribute closed designs in final form to end users, open technologies are developed through a collaborative and continuing process that blurs the distinction between developer and end user. The increasing convergence of ethical values between the media industry and the open movement suggests that as journalists incorporate sensor technologies into their work, they may be inclined to develop and use open designs. Legal and Ethical Concerns Associated with Open Designs While journalists may be drawn to open technologies, it is important to keep in mind that these designs may create more pronounced legal and ethical concerns than their closed counterparts. Before exploring the ways in which open technologies may be more problematic for developers and users, let’s outline the basic differences between the two.

Closed technologies are constrained by the applications and operating environments that are predetermined by their developers.5 In contrast, open technologies are intended for modification and are not limited by the applications or environments in which they are originally deployed.6 The distinct architectures of closed and open designs promote different liability outcomes and ethical considerations. Since developers of closed designs prescribe the applications and operating environments of their technologies, they are able to anticipate potential harms that may occur through the use of their products.7 Foreseeability of harm allows developers to incorporate safety features into their designs and to provide product warnings to end users. In contrast, since open designs are subject to modification and use in a variety of applications and environments, their developers are unable to anticipate potential harms that may occur downstream or to convey product warnings to users.8 Further, since developers of closed technologies prescribe permissible product applications and operating environments, they are able to rely on the defense of product misuse to shelter themselves from liability when their technologies are employed beyond the prescribed applications or environments. Conversely, since open technologies are intended to support an endless variety of applications and environments, it is unlikely that their developers could rely on this defense.10 Take for example, a media company interested in using drones to capture the news. For the purposes of this example, assume the company has only two concerns: 1) legal liability associated with bodily harm or property damage caused by the drones; and 2) an ethical concern regarding the use of the drones to violate privacy. If the company opts to procure closed drones, the liability issues are fairly predictable. The drones may be built with safety features, such as “sense and avoid” technology that reduces the risk that the drones will collide with people or property. If the safety features do not function according to their specifications, then the developer or manufacturer—not the media company— may be held liable for any personal injury or property damage that occurs. However, if the journalist operator modifies the drone or tampers with the “sense and avoid” system, then that operator—not the developer— will likely be found liable. If, on the other hand, the hypothetical media company decides to use open drones, liability becomes more problematic. Assume the journalist operator makes various modifications to the “sense and avoid” radar, turning it into a multifunctional system that also features communication and weather modes. If the revamped drone crashes into a person, causing bodily injury, who will be liable? A court would have to engage in a complicated and technical analysis to determine whether the underlying technology or the modified upgrade is to blame. Moreover, the initial developer of the open “sense and avoid” radar will not be able to avoid legal liability by claiming alteration as a defense. Should the hypothetical media company be concerned that its drones may be used to violate privacy, it may be inclined toward adopting closed designs that incorporate privacy-by-design features. Since closed technologies are not subject to end-user alteration, individual operators are not permitted to adapt the underlying technology, the result being that any privacy protections would remain intact. If a journalist tampers with a privacy feature, then the journalist would be in violation of the end-user license. In the

case of open technologies, modification is permissible; there is no mechanism that requires downstream users to maintain features incorporated by upstream developers. Possible Industry Measures Although from a liability perspective open technologies may be more problematic than closed designs, the good news is that a variety of industry measures may be adopted to mitigate liability risks and address ethical concerns. 1) Ethical Open License Developers of open technologies can look to licensing as a mechanism for allocating liability and promoting non-harmful and ethical use of their technologies. An ethical open license model is one way to prescribe tailored obligations and restrictions on downstream design and use. For example, an ethical license might prohibit end users from removing safety or privacy features incorporated by upstream developers. It may also prohibit the incorporation of technologies used to circumvent drone-defense systems12 installed by property owners seeking to protect their privacy from drones. By placing specific prohibitions on certain modifications or applications, upstream developers would have a means to shelter themselves from liability and to rely on the defense of product misuse. What distinguishes the ethical open license proposed here from typical open licenses is that it seeks to balance the goal of openness with other goals, namely the promotion of safety and privacy. Typical open licenses do not restrict downstream parties from making particular modifications or from using the underlying components in certain applications. The reason for this is that such licenses are preoccupied with protecting the ability of downstream developers to understand and modify the underlying technology. While this singular preoccupation is perfectly acceptable when software remains embedded in devices like personal computers, it must be tempered with the goals of achieving non-harmful and ethical use in the context of technologies like drones that have the ability to cause bodily injury and significant property damage.13 Open software licenses have typically emerged within three unique environments: academic (think the Berkeley Software Distribution [BSD] License and the MIT License), corporate (the Mozilla Public License and the Common Public License), and ideological (the General Public License [GPL] family). The style and substance of licenses generally reflect the environments in which the licenses have been developed. For instance, the GPL is presented in the style of a manifesto, and it is a restrictive copyleft license that is generally not compatible with closed proprietary offerings.14 In contrast, the Mozilla Public License (MPL) is drafted in a formal legalistic style, and is more permissive and suitable for use in proprietary offerings. Open licenses are generally managed and updated by the institutions with which they are associated, in conjunction with the broader open community. Take, for example, the revision process implemented by the Mozilla Foundation when it was preparing the second version of its MPL. As the custodian of the MPL, the Mozilla Foundation initiated a public process to engage users, lawyers, and the broader open source community to provide feedback on the MPL and offer suggestions for modification.15 The MPL 2.0 was the result of a revisionary process that lasted almost two years, during which several alpha and beta drafts were released for further comment.

Media companies thinking about developing drones and other sensor technologies in-house ought to consider creating their own open licenses that address liability risks and ethical concerns associated with these technologies. From time to time, such companies may employ a consultative model to update their licenses by soliciting engagement from the broader journalism community. 2) Media Industry Code of Conduct A complementary solution to the ethical open license is the development of an industry code of conduct to govern journalists’ use of sensor technologies. Similar codes have already emerged in other industries; for instance, The American Society for Photogrammetry and Remote Sensing (ASPRS) has developed an ethical code to govern the use of sensor technologies by professionals in the mapping sciences industry. The ASPRS’s code promotes the recognition of the “proprietary, privacy, legal, and ethical interests and rights of others.”16 It also requires subscribers to “not condone, promote, advocate, or tolerate” the use of sensor technologies “in a manner that knowingly contributes to… [the] transgression of reasonable and legitimate expectation of privacy.”17 Final Thoughts In the near future, drones and other technologies equipped with sensors will become more prevalent in the national airspace and it is likely that journalists and media companies will be among the early adopters. The ways in which the media industry develops and uses these technologies will doubtlessly impact public acceptance. Although open sensor technologies may be more problematic from a liability and ethics perspective, measures including ethical licensing and a code of conduct may be taken to mitigate liability risks and address ethical concerns.

By Mickey H. Osterreicher and Matt Waite The legal landscape in the United States for small unmanned aircraft systems (sUAS), weighing less than 55 pounds, is subject to changing winds as legislators and regulators from federal, state, and local governments consider how to integrate these devices into society and the U.S. airspace. Federal policy is being driven by Congress, drafted by the Federal Aviation Administration (FAA), and overseen by the courts. More than 40 states have considered restrictions—most dealing with law enforcement’s ability to use unmanned aircraft systems—while six have either passed or are considering laws that would directly impact journalists’ ability to use sUAS for reporting. While several cities have outright banned drones (as they are more frequently called) in their communities, the most significant impact upon journalists wishing to use sUAS in the United States will be determined by the FAA and the states. The FAA’s Current Stance As of this writing, the FAA insists that it has regulatory control over sUAS in the United States,1 and that only hobbyists and government entities with

explicit permission from the agency can fly them. Commercial operations are not allowed under FAA policy. The FAA’s primary concern is to protect the safety of other aircraft and people on the ground, however, Congress has recently requested that the agency consider issues of privacy when developing unmanned aircraft systems (UAS) regulation. In the FAA’s Modernization and Reform Act (FMRA) of 2012,2 Congress ordered the FAA to integrate UAS into the national airspace by September of 2015. The FAA, however, has been significantly delayed in meeting that schedule. The selection of six UAS test sites came almost two years late, and the procedure for sUAS rulemaking has been twice delayed. The FAA now says November of 2014 will be the earliest that it will make proposed regulations for sUAS public.3 Final regulations will take 12 to 18 months after that notice to complete. The FAA’s current policy is facing legal challenges. The administration fined drone pilot and promotional video maker Raphael Pirker $10,000 for a 2011 flight over the University of Virginia campus. Pirker’s lawyer, Brendan Schulman, filed a motion to dismiss and on March 6, 2014, a National Transportation Safety Board administrative law judge granted the motion by vacating the charge, setting aside the fine, and terminating the proceedings against Pirker with prejudice (meaning they may not be refiled).4 In the eight-page decision, Judge Patrick G. Geraghty wrote that none “of the definitions the FAA used for ’aircraft’ are applicable to, or include a model aircraft within their respective definition” and Pirker was “subject only to the FAA’s requested voluntary compliance” with rules for hobbyists. The judge further stated that the policy the FAA relied upon for its rules was “intended for internal guidance” and was “not a jurisdictional basis for asserting… enforcement authority on model aircraft operations.” Essentially, the court found that the FAA policy provided no power of authority over Pirker’s actions, and deemed the policy both non-binding and an invalid attempt at legislative rulemaking. While many viewed the decision as a green light for the use of sUAS, less than a day later the FAA filed an appeal,5 which immediately stayed the court’s decision and rendered the FAA’s current ban on commercial drone use unclear yet again— and its enforceability questionable. Other Legal Challenges There has been other pushback against sUAS operators beyond the instance involving Pirker. In February, police ordered Pedro Rivera, a photojournalist for WFSB TV in Hartford, CT, to cease flying his drone over the scene of a fatal car accident and vacate the area immediately. Officials later said Rivera was interfering with their investigation and contacted his employer, despite Rivera’s statement that he was operating on his day off. WFSB suspended the journalist for at least one week without pay. In response, Rivera filed a federal lawsuit against police, alleging violations of his civil rights. Rivera’s complaint states that “his device was hovering at an altitude of 150 feet” and that he was standing outside of the cordoned-off crime scene “operating his device in public space, observing events that were in plain view.” The complaint claims that “private citizens do not need local, state, or federal approval to operate a remote controlled model aircraft” and that police intended to “impede the exercise of” Rivera’s First Amendment rights to monitor the police response to a motor vehicle accident, while also chilling his free speech rights.

The FAA—Future Plans and Concerns Last November the FAA produced three documents,6 which when read together, provide a roadmap and expected timetable regarding future sUAS regulations. As noted in the introduction, the FAA’s traditional responsibility has been to ensure public safety, with the new added consideration of privacy. At the time of writing, the mechanisms the agency appears to be investigating for regulation purposes include pilot certification, restrictions for where sUAS can fly, and minimum hardware capabilities. However, to date the FAA has provided few specifics about what official regulations will look like, and less than a month after releasing the roadmap delayed its timetable again. With this in mind we can only speculate about what regulations may contain. The roadmap suggests that current FAA regulations requiring “pilot and crew certification,” which include meeting medical, training, and security/ vetting requirements as well as passing written and physical exams, could be applied to sUAS.7 Current FAA roadmap language refers to operation of sUAS in “approved airspace” and under specified “procedures.”8 Given the unpredictable nature of newsworthy events and the spontaneity of when and where they occur, the abstraction could be concerning for journalists. The FAA’s USA Comprehensive Plan, which includes portions of the roadmap, insinuates more specific and potentially untenable regulations restricting initial sUAS flights to “daytime”9 operations “not over populated areas.”10 Additionally, undefined references to “performance constraints” and “approved procedures” may also lead to arbitrary and capricious operational restrictions upon sUAS operations. Of equal concern may be a requirement that any sUAS be operated within the “visual line-of-sight (LOS) of the flight crew”11 or be subject to such things as filing a flight plan and using a transponder. Many of these potential rules would put regulatory compliance beyond the means of most sUAS operators. Additional issues that need to be properly addressed are vehicle registration and regulations regarding the recording and reporting of safety data. Aside from these safety concerns is the issue of privacy. Congress, in an Explanatory Statement to the Consolidated Appropriations Act, 2014,12 directs the FAA “to conduct a study on the implications of UAS integration into national airspace on individual privacy”13 and that the report “be completed well in advance of the FAA’s schedule for developing final regulations on the integration of UAS into the national airspace.”14 State legislatures, however, are not waiting for the FAA to act. State Laws In total, 44 states have considered or passed bills to restrict drone use, the vast majority of them aimed at law enforcement use. However, Idaho, Texas, and Iowa have passed laws that directly impact private and commercial use of drones. And, as of this writing, bills are also being considered in Washington, Wisconsin, and Georgia that would directly impact if, or how, sUAS can be used for newsgathering. For the most part, such legislative measures appear to address concerns regarding privacy rights and private property. In one form or another, the bills seek to either criminalize or provide civil penalties for photographing a person “where they would have a reasonable expectation of privacy” or make it illegal to photograph private property without consent of the property owner. Enacted: Texas: Texas law15 makes it a misdemeanor to capture images of a person or private property using an unmanned aircraft flying higher than eight feet in the air without permission of the person or property owner. The law also provides civil penalties for “disclosure, display, distribution, or other use of any images captured” in violation of the law—which allows nineteen exceptions under which such use is permitted.16 Newsgathering was excluded as one of those exceptions in the final amended bill. Idaho: Idaho law17 reads: “No person, entity or state agency shall use an unmanned aircraft system to photograph or otherwise record an individual, without such individual’s written consent, for the purpose of publishing or otherwise publicly disseminating such photograph or recording.” It exempts flying model airplanes so long as they are used for recreation or sport. The law does not criminalize such activity but instead provides “for a civil cause of action”18 and liability for damages and legal fees.

Iowa: The Iowa House bill19 criminalizes the use of an UAS “to capture an image of an individual or private property with the intent to conduct surveillance, or to stalk, follow, or intimidate another person”20 by making it an aggravated misdemeanor21 (punishable by up to two years in jail or a fine of up to $6,250). The companion bill in the Senate bars “entering upon or in property for the purpose or with the effect of unduly interfering with the lawful use of the property by others, including interference using visual, auditory, or photographic means that intrudes upon legitimate privacy interests in, on, or around private property that is not normally open to the public… .”22 If someone is hurt in the course of drone operations the charge is raised from a misdemeanor to a felony. Proposed: Georgia: The proposed Georgia bill23 is almost an exact duplicate of the Texas statute except it contains only eighteen exceptions, omitting as redundant the one that permits drone photography over private property if it is within 25 miles of the border with Mexico. Washington: In Washington, the proposed legislation24 intends to make the unauthorized use of unmanned aircraft equipped with a sensing device (including cameras) a “gross misdemeanor”25 punishable by up to one year in jail, a $5,000 fine or both.26 There are exceptions: drones may be flown above private property with the consent of the owner who has “received actual notice that the aircraft is equipped with a sensing device.”27 In this case the UAS must also be “clearly and conspicuously labeled” with the name and contact information of the aircraft’s owner and operator.28 Another exception is for drone flight “above public lands in a manner that does not unreasonably interfere with the rights of others and the operation is not otherwise prohibited by law or rule.”29 Wisconsin: Similar to Iowa, the Wisconsin bill30 “prohibits a person… from using a drone that is equipped with video or audio recording equipment to photograph, record, or otherwise observe another individual in a place where the individual has a reasonable expectation of privacy.”31 The penalty for such use is a Class A misdemeanor, subject to a fine of up to $10,000, imprisonment for up to nine months, or both.32 Constitutional Questions The First Amendment states in pertinent part that the government may not abridge freedom of speech or of the press. Through precedential case law, photography is deemed a form of expression protected as free speech. But those constitutional protections are not absolute. They are subject to reasonable time, place, and manner restrictions. Should regulations be put in place, they often cite the necessity for maintaining public safety. Such regulations must also be content-neutral and narrowly tailored while also leaving open alternative avenues of communication.

Applying these standards to a free speech/free press argument, the FAA may be successful in enforcing certain restrictions. But depending on the language it promulgates and the extent to which those regulations prohibit sUAS use, the agency may be open to constitutional challenges by journalists and others. Conclusion With the exponential proliferation of cellphone cameras and the ease with which images of all types (still and video) are disseminated, many legislators have attempted to regulate traditionally protected First Amendment activities with bills that have been dubbed as “anti-paparazzi” or “ag-gag.” The constitutionally suspect language of those bills is now being repurposed in attempts to prohibit the use of sUAS for various uses including newsgathering— even though most of the concerns the language addresses are already covered by the same common law tort principles used to allay privacy fears. States should expect more legal challenges against laws prohibiting the use of drones, especially in states where prohibitions are not content neutral, are overly broad and vague, and in the case of newsgathering, limit more speech than is necessary to achieve a significant governmental purpose. Most importantly, the lagging FAA rulemaking scheme and increasing state legislation emphasize the need for practical sUAS regulations that will strike a fair balance between concerns over air safety and privacy, and the rights of journalists and the public to use a new tool for newsgathering.

When Things Fail or Fall: Personal Injury and Negligence

By Deirdre Sullivan Deirdre Sullivan practices in the legal department of The New York Times Company. The views she expresses here are her own and do not reflect the views of The New York Times Company. Introduction and Scope Any responsible media organization or journalist operating or deploying sensors should consider the risks that their actions pose to other people. Neither regulatory oversight nor journalistic ethics will be able to fully account for human error or bad luck. With an eye to the component elements of a personal injury claim of negligence, there are a few practical steps that can be taken to anticipate and mitigate some of those risks. It is difficult to predict the outcome of any personal injury claim; where the technology is as young as it is here, and the precedents as limited, it is nearly impossible. Personal injury claims are among the most fact-intensive litigations, and, like unhappy families, each claim will be unhappy in its own way. No two are ever alike. The smallest details can be dispositive in a personal injury case, making the outcome completely dependent on case-bycase findings of fact. As a result, especially in personal injury claims, there is limited predictive value in hypothetical fact patterns. Still, hypothetical situations can be more than an academic exercise or a guessing game. If we agree that the law seeks to direct and constrain human actions and social behavior by thinking about the risk drone and sensor journalism pose to

people and property in the context of civil and common law, we can import the social values embodied by hundreds of years of common law to formulate a path forward. Most personal injury claims arise from an action. Sensors, by nature, are often passive devices—the action taken is most typically in the deployment and collection. The largely passive nature of sensors offers few obvious narratives to illustrate the key concepts of a personal injury claim. By contrast, a sensor deployed by a drone offers a practical, active example (i.e., a journalist is operating a drone with a sensor when it falls and strikes someone on the ground, causing physical harm to the person). This section often relies on that example. Keeping this in mind, we will start with a brief survey of the law’s longstanding attempts to manage the risks of things falling out of the sky, followed by a primer on the elements of a negligence claim within personal injury litigation, and end with an overview of some practical considerations for journalists or media organizations planning sensible and effective usage of drones and sensors. The Historical Perspective Drones and sensor journalism may be new, but the law has a long history of extending generous interpretation to the victims of falling things. The Digest of Justinian, a mid-sixth century codification of Roman laws, includes a section on “Those Who Pour or Throw Things Out of Buildings.” It sets out the then-current state of the law: If anything should be thrown out of or poured out from a building onto a place where people commonly pass and repass or stand about, I will grant an action to be brought against whoever lives

there for double the damage caused or done as a result. If it is alleged that a free man was killed by whatever fell, I will grant an action for 50 aurei.1 The Digest then clarifies that if something falls (as opposed to being “poured or thrown”) from above, liability remains because “it is in the public interest that everyone should move about and gather together without fear or danger.” On July 18, 1863, a barrel of flour plummeted from a second story stockroom belonging to Liverpool flour merchant Abel Boadle onto Joseph Byrne, a cork manufacturer who happened to be passing by at precisely the wrong moment. Byrne was incapacitated for weeks, left in a great deal of pain, and ultimately unable to earn a living for many months. Byrne hadn’t seen what had caused the barrel to fall onto his head, and neither had witnesses. He eventually sued Boadle for negligence and prevailed, establishing the important legal precedent of res ipsa loquitur along the way. The phrase—Latin for “the thing itself speaks”—informs the doctrine of negligence law, stating that a breach of duty can be inferred from the very nature of an accident without direct evidence of how a defendant behaved. On December 9, 1979, John Bowen, a 20-year-old from Nashua, N.H., was attending a Jets-Patriots football game in Shea Stadium (ancient history) when he was struck in the head by a “radio controlled model airplane” that was part of the halftime show. He died six days later from brain injury.2 Two years later, a $10-million damage suit was filed in Federal District Court in Brooklyn. Named in the suit were the Jets; the Radio Control Association; and Philip Cushman, who designed, built and provided the model airplane. The settlement amount was not disclosed.

Personal Injury and Negligence A personal injury claim against a journalist or a news media organization arising from sensor or drone journalism would most likely assert the tort of negligence.3 Most personal injury claims, including negligence, are common law claims. Common law is the law of precedent or custom (as opposed to written statute): A judge’s decision is made based on the holdings of previous cases, reflecting the belief that similar issues should be treated similarly. Through this elegant evolution, common law provides society a set of broadly applicable principles. However, the specific application of those general principles is a messy, fact-intensive, jurisdiction-dependent process. When an objective standard is brought to the chaos and complexity of human behavior, the outcome relies heavily on the finding of fact. Under common law, conduct is negligent if it subjects another to an unreasonable risk of harm arising from one or more particular foreseeable hazards, and a successful claim for negligence must demonstrate that negligent conduct caused the plaintiff damages. Today, most jurisdictions (and the Second Restatement of Torts) organize the requirements for a successful negligence claim into four elements: (1) Duty, (2) Breach of that Duty, (3) Causation, and (4) Damages. A plaintiff must prove all four elements for a claim to succeed.

Duty Put broadly, we all have the legal obligation not to place others in foreseeable risk as a result of our conduct.5 Regardless of the jurisdiction or statutory obligation, a journalist operating a drone to deploy a sensor has a duty to those around her to do so in a safe manner that does not present others a foreseeable risk of injury, much as one does when operating a car, driving a delivery truck, or riding a bicycle. The duty of care requires us to mitigate those risks. Breach If our conduct is unreasonable in light of foreseeable risks, we have breached our duty. The court evaluates behavior against the “reasonable person” standard, asking if a “reasonably prudent person” with the same duty would have engaged in the same conduct. Determining what this prudent person might do in a given set of complex circumstances can be difficult. It demands carefully structured reasoning, so much so that it’s spawned its own algebraic equation: B

6 For example, the difficulty of operating certain drones in a city might be known, and operating one in the wrong circumstance—perhaps without a GPS fix over an outdoor concert in Central Park—isn’t something the mythical reasonable person would be likely to do. Tweak the facts slightly, however, and the duty might not be breached: Perhaps there is some urgent safety need served that supersedes the known risks. In that case, a reasonable person might see the risk but proceed anyway, and the breach of a general duty of care might not actually have occurred. But if there was another, less risky way to address that need, then the calculus of risk shifts again. Causation In a personal injury claim, the plaintiff must also demonstrate that the injury was caused by the defendant’s negligent conduct: But for the placement of a sensor on the windiest side of a skyscraper, for example, the injury would not have occurred.7 A plaintiff can only recover for damages that were foreseeable and a natural consequence of the negligence. In the case of a sensor falling from the sky, for example, the victim is not likely to have witnessed the defendant’s operation of the device or, by extension, the defendant’s breach of her duty of care. (To the extent that personal injury results from the operation of a drone unseen by the victim, we can turn to the poor immortal Joseph Byrne and the case that established the doctrine of res ipsa loquitur in negligence cases.) Damages Compensation for the injuries caused by negligent conduct is at the heart of any negligence claim—indeed, it is the very purpose of the claim itself. If no physical injury or damage occurred, no damages will be required, and even if all the other elements of a negligence claim are made successfully, the claim will fail.

Practical Considerations Negligence Per Se in an Uncertain Legal Regime For the most part, action in violation of a statute (that is to say, a law), particularly a safety statute (e.g., running a red light or speeding), will conclusively establish negligent conduct. This is called negligence per se. If the placement or operation of a sensor is in violation of a law and injury occurs, the claim will succeed regardless of whether the injured person can show duty or breach. If the action is in violation of a local ordinance or administrative regulation (as opposed to a statute), however, courts are split on whether to find negligence per se or to continue with a full negligence test. Applying the split to the current uncertainty around FAA regulation of U.S. airspace, and the inconsistent laws across states, there is a reasonable chance that merely operating a drone for commercial purposes could establish negligence per se in one jurisdiction but not another, and could theoretically make an operator’s training and certification an important fact—again, in one court but not another. Insurance and Litigation In a personal injury lawsuit, everyone up and down the chain of events should expect to be named in the lawsuit—the journalist in her individual capacity; the drone spotter; the media organization; the manufacturer; and the individual responsible for the maintenance, testing, and calibration of the drone or sensor. Insurance can mitigate exposure to a negligence claim and will probably become a key consideration for any news organization or journalist engaging in sensor journalism, especially where sensors are deployed using drones. In the near term, we can expect insurance companies to enter cautiously into the commercial drone market and explore opportunity with narrowly scoped policies. Premiums for a drone operations policy may be relatively

high until underwriters have data points (in the form of real life examples) to more carefully calibrate the spectrum of risks they would be underwriting. High premiums may present a material barrier to entry for smaller newsgathering organizations and freelancers who are only occasional users. Most large news organizations have a general commercial liability policy that may cover personal injury arising from use or deployment of sensors. However these policies typically contain an aircraft exclusion that would not cover claims arising from employees’ operation of drones. Even corporate- owned aircraft insurance policies are unlikely to address the expected risks presented by drones. Freelance journalists who own and operate their own devices may not be protected by the terms of their engagement and would likely want to get their own insurance, but could again be confronted by high premiums. Even if a commissioning publication indemnifies the freelancer, that indemnity would probably not include damages and liability arising from the freelancer’s negligence. If insurance is prohibitively expensive, or the use of drones is only occasional, organizations and journalist may prefer to outsource exposure—that is, hire a third party to operate a sensor drone. Risk is bought and sold every day in service agreements in the form of indemnification and insurance provisions. If the operation, maintenance, and calibration of drones or sensors force a journalist or organization to confront unwelcome questions of potential negligence, contracting with an experienced and qualified third party willing to provide indemnification (though possibly at a premium) might be a useful alternative to an in-house drones or sensor program.

Journalism and Truth in Sensing

By Lucas Graves Introduction Questions of evidence and truth are at the center of journalism, and also strangely invisible in day-to-day news routines. Reporters sometimes speak as if the truth “is something that rises up by itself, like bread dough,” write Bill Kovach and Tom Rosenstiel in The Elements of Journalism. “Rather than defend our techniques and methods for finding truth, journalists have tended to deny that they exist.”1 In everyday reporting, the problem of sourcing can mean anything from locating the documents that will make or break a story, to finding someone willing to be quoted saying what everybody knows to be true. While Bob Woodward and Carl Bernstein’s book, All the Presidents Men, made the two-source rule part of the basic vocabulary of American journalism, in practice, standards of evidence have varied widely from outlet to outlet, and from story to story. (Even at the Washington Post of the 1970s the standards of the Watergate investigation were an exception, not the rule. At least one Post editor has said the two-source requirement was a myth.)2 And many reporters believe that standards of evidence have eroded in the “hamster wheel” of contemporary journalism.3

Around-the-clock news on cable and the Internet have quickened the news cycle, and invited formats that defy tight editorial control. The practice of holding back untested claims can seem not just infeasible, but antiquated, as professional journalists cede their gatekeeping role. When applying this trend to sensor journalism, upon first consideration the turn toward data- and sensor-driven reporting seems to ease some of these tensions. Reporters who gather raw data about the social or natural worlds appear less dependent on human sources, less vulnerable to manipulation or to their own biases, and less tempted to fall back on “he said, she said” story construction. The kinds of questions asked and answered in sensor journalism seem to fall outside personal interests and ideologies. In practice, of course, data is never raw, and sensors don’t speak for themselves. Reporters working on these kinds of stories may depend on experts, not just to interpret results, but also to conceive and plan investigations. Most important, errors, biases, and assumptions can be trickier to account for as they are pushed farther upstream into the design of field tests, the calibration of sensors, and the definition of what counts as data. Sensor Evidence and Reporting Work The field of sensor journalism is new and open enough to encompass reporting projects that are otherwise very different, from aerial surveillance with unmanned drones, to measuring the shifts in soil temperature that awaken seventeen-year dormant cicadas. One useful starting point is identifying the use of evidence in stories built around various ways of monitoring the human and natural environment; then investigating the role sensed data plays in reporting and story construction. A sensor may be a hardware device monitoring changes in water levels in a Wisconsin stream, or noise levels in New York City. It may be a simple computer script tracking volumes of BitTorrent traffic, for instance among Academy Awards Best Picture contenders. In each of these cases the sensor acts as an instrument for transforming sensed information into a predefined category of data used

to buttress reported claims—about the environmental effects of frac-sand mining, for instance, or socioeconomic patterns in noise pollution, or even a correlation between Oscar nominations and illicit online film swapping. In other words, the sensor testifies to facts about the world, which in another story would rely on some human or institutional authority, or on the reporter’s own observation. Usually the sensor is tailor-made to offer a very particular kind of testimony and nothing more. In this vein, sensors share important characteristics with one tool with which reporters are much more familiar: the opinion poll. In the United States, newspapers and magazines helped to pioneer opinion polling in the early 20th century; independent pollsters such as Gallup and Roper began to develop more rigorous techniques in the 1930s. Polls are not sensors, of course. A CBS News/New York Times phone survey run out of a North Carolina call center is not automated in the same way as a soil temperature gauge. But from the reporter’s point of view the poll and the thermometer do a similar sort of work. Each transforms some aspect of the world into newsworthy data. Structurally, a front-page story on the latest Quinnipiac survey may have less in common with other campaign-trail coverage than it does with a piece reporting on environmental data. In both cases, the numbers themselves often supply the news value and the headline. Interviews and other reporting work serve mainly to interpret or contextualize those findings. And wellestablished categories measured regularly over time—whether consumer confidence or mean annual temperature—become story genres in their own right, deserving coverage whenever the latest data becomes available. These examples underscore the link between sensor journalism and the wider turn toward data-driven reporting, today and in the past. The lines between sensing data, mining data, and analyzing or interpreting data can be difficult to draw. Consider two recent landmarks of data journalism: the 2005 series on Florida’s disappearing wetlands in the St. Petersburg Times, and the 2011 reports by San Francisco’s Center for Investigative Reporting (CIR) on substandard school construction in California’s seismic dan ger zones. In an obvious way these reports relied on data from two of the most advanced, and expensive, sensor networks developed over the 20th century: orbital imaging satellites and seismographic monitoring stations. But in each case, obtaining data from these sensors marked only the beginning of a long and difficult reporting process. The St. Petersburg Times (now Tampa Bay Times) investigation began with the straightforward idea of using software to analyze and compare satellite imagery from 1990 and 2003 to show exactly how many acres of Florida wetlands were being lost to development—a figure neither the state nor federal government could supply. “I went into it with a simplistic idea of glorified subtraction,” the story’s co-author Matt Waite explains in an interview. In the end the analysis took ten months. The satellite images were handcoded into 100 separate pixel groups, based on three bands of electromagnetic spectrum that are especially sensitive to moisture in plants, and then compared using a custom-built algorithm. This yielded the shocking figure of one million acres lost—until Waite realized he had made a mistake: The tide was lower in the later image. “I had just wasted tons of time and money to show when the tide was out,” he recalled. “I thought I was fired.” But drawing on other datasets allowed the investigation to continue. The final published analysis revealed 84,000 acres of wetlands lost, based on reconciling the satellite imagery with data such as land-use records and maps of urban land cover. The results were further verified against higher resolution aerial photographs, available for some areas, and by sending reporters out to visit a random sample of individual sites. Three professors were asked to review the analysis before the story went live, and the paper published a detailed methodology online. The “On Shaky Ground” series at the Center for Investigative Reporting unfolded in a similar fashion. The investigation began in 2009 when reporter Cory Johnson, working on a story pegged to the 20th anniversary of the Loma Prieta earthquake, placed a records request with California’s Division of the State Architect (DSA), which enforces safety standards in school buildings. To his surprise, the DSA returned a spreadsheet listing more than 9,000 out-of-compliance schools. This launched a nineteen-month effort by CIR to build a comprehensive database of unsafe schools in seismic hazard zones, drawing on three separate datasets: the DSA’s school safety records, the index of school building locations maintained by the Department of Education, and seismic data from the California Geological Survey. As in the wetlands investigation, messy official data—misspelled school names, inaccurate building locations, and even shifting definitions of seismic hazards— had to be painstakingly cleaned up with traditional reporting tools like interviews and site visits. An academic analysis of the investigation, by sociologist Sylvain Parasie, found that the journalists involved came to think of building the database the same way they thought of writing a story: as a “reporting process” subject to the usual techniques and judgments of journalism. Once the team had confidence in the database, though, the reporters could use it to suggest new story angles and “drive” their investigation.4 These two investigations hinged on matching sensed data, from satellites and seismic stations, with other public databases. Crucially, in each case it was correlations across these datasets— buildings where wetlands used to be and schools with poor safety records near a fault line—that produced the key reported findings. This suggests an interesting conclusion: The most important sensors, from the point of view of reporting work, were the algorithms that looked for patterns in the data. In both cases these algorithmic sensors required constant fine-tuning based on information supplied by traditional reporting tools. The journalists’ confidence in these stories ultimately came from the work that went into cleaning up messy data and calibrating the algorithms used to interpret it. “In the end we felt like we could be pretty declarative about it,” Waite says. “The night before a big exposé you’re usually worrying about whether your sources will hold up and what the reaction will be. The night before the wetlands story I slept like a baby.”

Objectivity and Interpretation in the Age of Big Data Seen in this light, sensor journalism falls into a long history of efforts to build sophisticated, computer-driven analysis into reporting work—what in the 1970s came to be called “computer-assisted reporting” (CAR). The lure of CAR and so-called “precision journalism” rests at least partly with the promise of a kind of news that depends less on subjective interpretation and unreliable human sources. As Philip Meyer argued in his 1973 guide, Precision Journalism, “We journalists would be wrong less often if we adapted to our own use some of the research tools of the social scientists.”5 Hard data and rigorous methods would pull the reporter onto firmer evidentiary ground—something visible in the signal examples of this kind of journalism, and in the two cases described above. It is hard not to hear echoes of that promise in the contemporary rhetoric of big data in journalism, just as we have in the fields of business, government, and academia. As advertising campaigns like IBM’s “Smarter Planet” make clear, the big data revolution is being fueled by sensors of every imaginable kind: medical, environmental, industrial, urban, and so on, from Fitbit bracelets to RFID tags on shipping containers. This ubiquitous real-world surveillance comprises the much-hyped “Internet of things,” erasing the barriers between offline and online networks. The unprecedented volumes of data available from sensing threaten to make the scientific method itself obsolete: “This is a world where massive amounts of data and applied mathematics replace every other tool that might be brought to bear,” former Wired editor Chris Anderson wrote in a 2008 article. “Out with every theory of human behavior, from linguistics to sociology… With enough data, the numbers speak for themselves.”6

This sort of breathless pronouncement is easy to lampoon, but it reflects a general sense that relentless measurement of the world around us has made new kinds of objective insights possible. In a 2012 article, media scholars Danah Boyd and Kate Crawford called this the “mythology” of big data, “the widespread belief that large datasets offer a higher form of intelligence and knowledge that can generate insights that were previously impossible, with the aura of truth, objectivity, and accuracy.” They warn of the tendency for even academics to behave as if questions of interpretation belong strictly to the humanities and other “qualitative” domains, while “quantitative research is in the business of producing facts.”7 Scientists share these worries. In a recent editorial, Mike Fienen, a research hydrologist with the U.S. Geological Survey, argues that “in an era of big data” it is all the more important for scientists to be frank about the subjective interpretations and judgments involved at each stage of research, from designing experiments to reading the results. “Should we ignore this veiled subjectivity and pretend that cold hard data provide all the information we need?” he asks. “Or is it time to recognize and embrace the value of subjectivity?… It would be foolish to ignore the data, but it may be equally foolish to assume there is only a single way for the data to ’speak.”’8 Like other strands of environmental research, hydrology has made eager use of ever-cheaper sensor technology. Fienen models groundwater systems and is involved in a just-launched, five-year effort to track groundwater changes in Wisconsin’s Chippewa county with a network of monitoring wells and stream gauges. These measurements are vital to understanding the potential environmental effects of frac-sand mining, and Fienen works closely with journalists covering the controversial topic. “It’s an important question and there’s a big vacuum of information,” Fienen explains in an interview. At the same time, he adds, “It can be really hard to interpret the data. I deal with a lot of uncertainty, with conveying uncertainty in my research, and it’s always a challenge.” Journalists reporting on sensed data must appreciate three separate sources of uncertainty, according to Fienen. The most obvious, but generally least important, is the degree of accuracy attached to the sensing equipment. More difficult to grasp is what an accurate sensor reading actually represents. “Is it telling you what’s happening in one square centimeter of ground, or across the entire state?” Fienen asks. That depends on the number and distribution of sensors but also on the scientist’s model, a third major source of uncertainty. Scientists interpret their data according to a model of the world or of society, even as they try to improve that model based on an intuitive sense of what the results should look like. Subjective judgment always comes into play, whether reporters deploy their own sensors or rely on experts to provide data. What is vital to understand is that in practice, for reporters as for scientists, the subjective judgments and hands-on work involved in producing reliable data and making it meaningful become a source of confidence, rather than one of doubt. This suggests that the right way to think of sensor journalism is not as a source of self-evident facts and hard-coded mechanical objectivity, but rather as a continuation of mainstream reporting’s five-decade shift toward a more interpretive, analytical, “meaning-making” style of news.9

Technical Considerations in the Design and Execution of Sensor Journalism Campaigns

By Beth A. Stauffer Introduction The Internet of Things is changing how people interact with their surroundings, the environment, and each other. New technologies are allowing anyone with a question to attempt answering it using low-cost sensors, online datasets, and the power of cloud computing. The use of such sensors and information by journalists comes with slightly higher expectations for accuracy, reliability, and credibility. As a result, journalists must develop a sufficient understanding of how to design a data-collection campaign, choose appropriate tools for their questions, and assess data for quality and consistency. For scientists engaged in basic research, questions about experimental design, data quality, and reproducibility are typically dealt with in the peer-review process and through the copious use of statistical analyses. For nonprofessional participants in the scientific process, upfront planning and development of quality assurance plans can help identify the relevant sources of error and highlight effective ways of controlling for data quality. This essay is meant to serve as a guide to the minimum requirements journalists should consider when pursuing data collection and/or including analysis and interpretation of existing datasets in their work. In addition to assuring that a campaign is well-designed or that an existing dataset is

of high quality, a small investment of time in the initial stages can save a significant amount of effort down the line in the interpretation and communication of credible results. Sampling Design—How to Design for Answers There are potential sources of uncertainty at every step along the way in the collection and analysis of data. These must be considered both in the design of new sensor-driven data collection campaigns and in efforts to extract meaning from existing datasets. Sources of uncertainty can be encountered in the design of sampling campaigns and choice of measurement technique, input of data to centralized repositories, and analysis and interpretation of data by both providers and end users. While minimizing uncertainty at all stages is the ideal approach, proper documentation can also serve the purpose of providing context for data that might be otherwise scrutinized into oblivion. There are a few basic principles that should be considered when designing a sensor journalism campaign. These principles largely provide direction for how often and where in space samples should be taken.1 For phenomena changing over time, signal-processing theory suggests that you should sample at least twice as often as the scale of time you’re interested in understanding. For example, if you think air quality is worse in summer than in spring, you need to make measurements every one-and-a-half months to resolve this seasonal variability (approximately three-month timescale). Recent criticism of an effort to map New Yorkers’ running routes using publicly available RunKeeper data2 has pointed out that the routes seem heavily biased to the period of time around the New York Marathon and did not take into account seasonal variability in outdoor recreation.3 Sampling on inappropriate timescales runs the risk of missing critical data and aliasing signals relative to other processes. On the spatial scale, it is important to be systematic when determining sampling locations. The U.S. Environmental Protection Agency (EPA) suggests randomly selecting sites that are representative of the area you are trying to study (probabilistic design).4 This approach can be especially important when little is known about the study area prior to beginning the campaign and when primarily investigating broad trends over space and/or time. However, to highlight disparities between locations, sampling schemes that employ a combination of selected sites and control sites beforehand can be powerful. For example, an investigation into how well green infrastructure projects contribute to overall improvements in water quality in a lake or stream should use probabilistic sampling since a spatially distributed, longterm trend is the objective. Conversely, The Washington Post maps gun violence in the District of Columbia using data from the ShotSpotter program in which gunfire-detection sensors are concentrated in parts of the city with the highest rates.5 Choosing the Right Tools—Limits of Detection, Accuracy, and Precision Once you have decided on a sample design to generate the most easily interpreted data, you must consider how you will actually collect data. From approved and established methods, to innovative, DIY, sensor technologies, the possibilities are numerous. Most sensors and methods come with some set of technical specifications, which often include the lower and upper limits of detection (LODs). These limits, however, are unfortunately not the straightforward edicts that they would seem; rather they are often based on statistical analyses of different types of measurement error.6 Both the upper and lower LODs also often represent the best case scenarios obtained by technical personnel under highly controlled conditions with properly calibrated, maintained, and operated instrumentation. Taken together, these caveats further underscore the need to choose an appropriately conservative tool for the ranges you expect to encounter and to try to validate performance over time.7 In addition to LODs, accuracy is typically presented as a technical specification for a given technique, methodology, or instrument. Accuracy is an estimate of how close a measurement comes to a known or standard value. This differs from precision, which indicates how close repeat measurements of the same sample come to one another. It is possible to be both inaccurate but precise and accurate but imprecise; however, the goal is to be both.8 Accuracy may be reported as either units or a percentage of the reading for the measured parameter. For example, a commercially available laboratory thermometer with a stated accuracy of 0.5 degrees Celsius across the range of -10 to 70 degrees Celsius should yield values between 14.5 and 15.5 degrees Celsius for a sample that is 15 degrees Celsius. Other instruments may use a percentage of the reading to communicate expected accuracy. Again, it should be kept in mind that accuracy and precision specifications are only estimates given proper maintenance, calibration, and use of instruments and methods for measurement. The need for highest accuracy and precision will necessarily scale with the intended use of the data, a concept formalized in, for example, Ohio EPA’s “3 Levels of Credible Data.”9 That is, for uses of data requiring legal defensibility, levels of accuracy and precision should be consistent with approved methods for measurement. The EPAapproved Hach Method 10360 for measurement of dissolved oxygen in water, for example, requires initial accuracy and precision of 5 percent and 2 percent, respectively.10 These specifications alone do not ensure legal defensibility, and dissolved oxygen data with lower levels of accuracy and precision (e.g., 5–10 percent) may still be appropriate for other, less rigorous uses. There have already been examples of good and bad tool choices in sensor- and data-driven journalism. In the RunKeeper example previously highlighted, an additional source of criticism was around the “fuzziness” of the location information.11 While smartphone-based units are surprisingly decent when compared to more expense GPS units, their ability to detect your location down to a few meters while running quickly diminishes with reduced access to clear sky,12 a significant issue in a metropolitan city such as New York. On the other hand, journalists in Florida enlisted the help of a well-known GPS company to provide tools with the specifications necessary to accurately locate interstate tollbooths in their investigative report on speeding cops.13

Metadata—The Benefits of Reconstructing the Who, When, Where, and How When collecting your own or considering whether to use data available from existing programs, it is always advisable to find out as much information as possible about techniques, equipment, and reporting standards used to collect the data. Often described as “data about data,” this metadata is essential to understanding, locating, using, and managing data.14 Metadata can be as simple as recording who took a sample or measurement, weather conditions at the time of sampling, and any interesting observations or oddities noticed at the time and place of sampling. Additional information about what equipment was used to make the measurement, how recently an instrument was calibrated, and any issues during the measurement/analysis process can also add value. Collection of metadata is an integral part of many assurance project plans (QAPPs), which many nonprofessional participants in science develop with federal agency guidance and have approved by local agencies. This approval process combined with the transparency of the protocols themselves, lends a higher degree of credibility to programs and the data they produce. Such plans ensure a high level of data quality throughout the design, collection, measurement, and analysis processes. Methods for data quality assurance can range from personal knowledge of a participant’s skill/expertise to algorithm- or statistics-based automatic filtering of data as it is reported.15 While more subjective approaches for data quality assurance might be sufficient for volunteer programs, ethical considerations and concerns about defensibility in sensor journalism may require using datasets with more robust filtering approaches built into the data quality assurance process.

Together, this metadata can help trace anomalous data in datasets and allow for attribution to real differences versus errors in collection, measurement, or analysis of samples. Leveraging Experts—How to Collaborate for Maximum Results Currently, few journalists are scientists, statisticians, or theoreticians. Over their careers journalists may investigate a range of stories that specialists have spent years researching. Given that, in most cases, they cannot be expected to obtain the advanced degrees to be able to responsibly and credibly use sensors and data in their work. Partnering with trusted experts or legacy institutions, which can include state, local, and federal agencies; universities; and colleges; can facilitate the identification of important considerations when designing a sampling program and provide valuable validation for techniques, instruments, or methods used in analyses. An excellent example of such a partnership can be seen in the Alliance for Aquatic Resource Monitoring (ALLARM) program between Dickinson College and Trout Unlimited, a nonprofit organization dedicated to conserving coldwater fisheries and their watersheds.16 ALLARM provides training, technical, and analytical support for Marcellus Shale water monitoring programs run through the Coldwater Conservation Corps of Trout Unlimited, and in turn the volunteer monitoring program receives higher quality data than possible through their limited in-house capabilities. While the relationship between scientists and journalists has not always been one of mutual respect and cooperation, scientists’ increased focus on skills for communication with the media and public and journalists’ interest in sensor- and datadriven work may provide an opportunity to rewrite this relationship.

Conclusions Journalists planning to collect their own or incorporate existing data into their work should utilize the wealth resources available for other nonprofessional endeavors in scientific research. These resources include valuable guidance for choosing the most appropriate sampling designs and tools, what to include or look for as metadata, and considerations for verifying data quality throughout the process. However, as journalists, the ethical and legal implications for data use are likely higher than for many citizen-science or volunteer monitoring programs. Sensor journalists should therefore strive to also consider requirements for legal defensibility in designing their campaigns or judging the quality of existing datasets. By keeping these considerations in mind, journalists should be able to confidently and ethically collect and incorporate data-driven material into their work.

The People Within the Pixels

By Lela Prashad Only recently has remote sensing become a viable tool for journalists looking to obtain images of events as they happen on the ground in near real time. Sensors that capture images from satellites, airplanes, and drones are now commonly used to visualize the effects of people’s behavior on the ground, rendering conflicts, urbanization, deforestation, and pollution in a new light. Most remote sensing used in journalism today is visible imagery, relying on sensors that see the Earth’s surface in the same colors that our eyes do but from a higher altitude and wider field of view. A few pioneering journalists, however, are beginning to use even more sophisticated remote sensing data that measures more than what the eye can see. As journalists begin to integrate remote sensing into their reporting, they are encountering many of the same ethical considerations government and academic researchers have in the past, especially regarding security, privacy, and access to data. The United Nations Platform for Space-based Information for Disaster Management and Emergency Response (UN-SPIDER) is charged with ensuring that imagery is shared promptly after a natural disaster for humanitarian assistance through the International Charter on Space and Major Disasters; the United States Agency for International Development (USAID) Famine Early Warning System has worked tirelessly to reduce complex remote sensing into simpler tools and put imagery into the hands of decision-makers around the world. As noted above, access to data is key. In the past, privacy and security were less scrutinized since very high-resolution imagery was not easily available. However, as technology improves, we must revisit some remote sensing experts’ theories that imaging people’s behavior on the ground, even at relatively course scales, has ethical concerns. While remote sensing provides valuable data, often unobtainable from other sources, imagery on its own displays a significantly incomplete picture, especially when used to understand people and their activities on the ground. Though an image is a single snapshot taken at one point in time, it contains discernable features and objects and can appear to tell a complete story. In order to fulfill a narrative, though, we must integrate imagery with what we’ll call ground-truthing—supplementing the image with other data sources and information to link its pixels to environmental phenomena and human activity. Remotely sensed imagery documenting the environment has been collected since the 1820s when Joseph Nicéphore Niépce captured the first permanent images of nature. Today, sensors mounted on airborne vehicles can offer images that replicate the way we perceive the world, but with additional energy and color beyond what our eyes can see. Outside of our familiar redto- violet spectrum of visible light, sensor instruments render near-infrared light reflected strongly from chlorophyll in plants on the ground. Nearinfrared imagery can be used to monitor forest and crop health in ecological studies, while instruments that capture thermal infrared can also map the composition of materials on surfaces and the patterns of heat they create— including the building materials of our cities. Still other instruments use active sources, such as radar, to broadcast signals to Earth that return imagery through the clouds, trees, or ground to reveal hidden fault traces under dense canopy and archeological ruins below the Earth’s soil. Until recently, these sensors have been prohibitively expensive for people outside of government and private companies to design and launch on their own. While NASA, the National Oceanic and Atmospheric Administration (NOAA), and the U.S. Geological Survey provide most of their sensors’ data for free, this information does not usually come in formats that are easy to

understand or use without advanced technical knowledge. Other countries outside the United States do not always freely provide their data, and private companies rarely do so either. Today, low-cost, open source initiatives are aiming to put remote sensing technology within reach for anyone interested in obtaining sophisticated imagery from balloons and kites, UAVs, and small satellites. Advocacy groups, like Public Lab; do-it-yourself (DIY) science and technology hobbyists; and a few intrepid journalists have started to utilize these sensors to mine for public knowledge. While these sensors are still far from the quality and coverage of flagship science satellites, such as Landsat, ASTER, or MODIS, their potential is quickly growing. A distinct advantage of satellite imagery is that is captures a synoptic view of the Earth’s surface—without regard for political boundaries—and outputs measurements that are equivalent and comparable no matter where on Earth they are acquired. Ground-based observations are often restricted to borders and boundaries where local governments or organizations have obtained funding and permission to install monitoring equipment. Comparing observations and measurements between two locations on the ground can be difficult, even within a single region, if different methods are used. There are currently many places on Earth with very limited or no ground-based measurements. Even imagery obtained from airplanes has the limitations of ground-based sensors, as planes must fly repeatedly over the same area to build up an archive over time to track changes and do not offer global coverage since they are not in orbit around our planet. No study that utilizes imagery is complete without ground-truthed local data, and this is especially true for studies focused on people and communities. One of the most common kinds of remote sensing products that attempts to image human activity on the Earth is land-use classification. Land-use classifications utilize imagery from satellite and airborne sensors to categorize imaged pixels into different categories. A simple land-use classification might distinguish urban, rural, and water-covered areas by identifying unique signatures of each type. Classifications have many advantages— they can be automated and run across archives of imagery acquired

over decades; they usually employ the same algorithm in any image so that a classified pixel of urban scape in New Delhi is calculated in the same manner as an urban pixel in New Mexico. Much more detailed classifications can be created to study urban growth and change over time, or loss of biodiversity and urban green space. While land-use classifications are ubiquitous and widely used across physical and social science studies, there is not a single gold standard classification for all purposes. The concepts of rural and urban are defined differently by individuals and disciplines and are always in flux; remote sensing cannot address questions of people and cultural institutions. However, these classifications are not problematic when their limitations are understood. In fact, they are indispensable when comparing phenomena across many regions that fall within a wide range. Efforts to map slums with remote sensing imagery highlight the need to integrate information about the people being mapped within the pixels. Governments, the United Nations, and NGOs use this type of imagery to detect slums by automatic measure and map their extents. Slums are typically undefined, informal settlements, and mapping boundaries and population density can help researchers understand the extent of assistance people living within them may need. There are significant ethical questions about mapping people from above, however, especially when it’s without their input—even when the maps are used to advocate on their behalf. Not all of these mappers share the residents’ interests. Many slums around the world encroach on valuable property and local authorities may be interested in using these maps to identify them and remove their residents. Developing automated classifications of slums does not lead to understanding how people live within these pixels, how they make a meaningful life with severely limited resources, or who they are as individuals. Imagery, however, can provide important tools for residents of slums and their advocates to understand environmental hazards plaguing the areas, such as air and water pollution, flood and landslide risks. Mapping communities for empowerment and advocacy has been an effective way for residents of marginalized com munities to affect change. Slum/Shack Dwellers International and Map Kibera are strong example organizations that seek to put remote sensing and mapping tools into the hands of those who are often mapped remotely by others, allowing them ownership of the information they collect. It can be easy to see the ethical implications of imagery that attempts to map and classify people and their behavior. It is difficult to discern the indirect impacts other kinds of environmental remote sensing have on people’s lives. Ecologically, remote sensing has been an extremely valuable tool in measuring and tracking deforestation around the world, especially in places where people are cutting and burning forests. This imagery is effectively used to estimate the loss of biodiversity, loss of biomass that can absorb greenhouse gases, and the generation of these emissions from burning forests. It’s now easy to track where local community members living in or near forests in environmentally sensitive areas are effectively destroying trees and landscape—though only the impacts of their activities can be imaged, not the reasons. The economic, cultural, and social pressures that are contributing to deforestation are as important to understand as the extent and environmental impact of deforestation itself. It’s clearly not possible to image these human factors directly; they require a very different kind of ground-truthed information. The necessity for investigating the people behind imagery is becoming clear to those seeking to affect change. The TREE Foundation project to protect the church forests of Ethiopia, founded by ecologists Dr. Meg Lowman and Dr. Alemayehu Wassie Eshete, seeks to address these issues. Satellite remote sensing data shows that nearly the entirety of northern Ethiopia has been deforested by local people planting crops for subsistence farming. Small islands of forests have been preserved by ancient Coptic Christian churches who hold the forests around the churches to be sacred. The TREE Foundation and partner scientists have conducted studies with satellite remote sensing and measurements made in the forests in partnership with community leaders in the churches. The church communities are participating in

the research and organizing efforts to build walls to preserve the renaming forests, which serve as islands of biodiversity for plant, insect, animal, and bird species. In the United States, a crowdsourced storytelling effort called iSeeChange is bringing together citizens and scientists to document the trends people see in weather patterns and how climate change impacts their daily lives. In the fall of 2013, a NASA Databridge workshop was held in Pasadena, Calif., to discuss how to integrate NASA remote sensing data and sensor design more closely with people and their activities. The workshop was facilitated by iSeeChange and representatives of NASA’s new Orbiting Carbon Observatory satellite. This communication between scientists, engineers, and the public represents the future potential of making imagery more understandable and accessible to everyone. The barrier to entry is rapidly lowering for those who want to design and launch sensors on airplanes, drones, and satellites and those who seek sophisticated types of imagery that extend beyond what the human eye sees. Journalists have a powerful new potential to integrate remote sensing imagery into their reporting and connect to a near-real feed of data for environmental issues, conflicts, and people’s activity on the ground. While imagery may appear to represent a full story, an image, no matter how sophisticated, is only one data point taken at a snapshot in time. Remote sensing imagery must be integrated with other ground-truthed data, local knowledge, and a firm understanding of people and communities to pull together a full story. Journalists, researchers, and others seeking to derive knowledge and understanding from sensors should see imagery as both a powerful tool and limited facet of a human story taking place within the pixels.


The Fourth, Final Section: Recommendations


Recommendations Journalists and news managers convinced by the case studies and arguments made in this report can take actions to better integrate sensors into their reporting. These range from adopting certain work practices, to making strategic long-term moves, to taking collective action as an industry. These recommendations are intended to take advantage of opportunities and stave off potential threats. Strategic Actions These are the general moves that forward-thinking journalists and newsroom managers can make to position themselves to benefit from the use of sensors as reporting tools. Identify and cultivate sensor sources for the beats you’ve prioritized. Many newsrooms can predict the important topics, stories, or themes they plan to cover over the next one to five years. In some organizations these might include local beats, in others they will be a set of subject matters. In either case, newsrooms should benefit from investigating whether there are ways to incorporate sensors into their reporting. In a small set of instances the sensors will already be in place and the data they produce will be easily accessible, understandable, and directly applicable to the specific story the newsroom is reporting.

In more cases, however, sensors may exist, but journalists will have to work hard to get the data or overcome that data’s limitations for journalistic storytelling. For other stories, it may even be feasible for newsrooms to build or commission their own sensing systems to gather data. In the majority of those cases, when the process is likely to be complex, newsrooms should partner or recruit experts to help them get the story right. It will take most newsrooms time and effort to cultivate their ability to work with sensors, both in terms of producing data, accessing it, and learning how to interpret it. However, the case studies show these kinds of efforts have paid off in the past. Newsrooms with access to new sources of sensor data and the expertise to interpret it for their communities may enjoy a significant competitive advantage. Put a watching brief on open source sensing systems. When it comes to widespread utility in the newsroom, open source and DIY sensing hardware is nowhere near the realm of open source software. Also, open source software’s wild success may not necessarily predict the same results for hardware. As noted in the case study about Public Lab, open source hardware designers fight headwinds that software designers do not. However, the community is active and committed. If, over time, open source hardware only provides newsrooms a fraction of the benefit derived from Python, R, and countless other open software projects, that still equates to a lot of value. Newsroom managers would be wise to make sure someone within their team is keeping an eye on open source and DIY sensing hardware, and continually assessing whether it can meet their journalists’ needs. News nerds should do hardware too. From the previous two recommendations, it follows that news organizations can gain advantage by incorporating sensors into their collective expertise. Organizational priorities will dictate whether those specific skills will be

building, interpreting, commissioning, or accessing sensors. Whether most newsrooms concentrate technical skills in a group of specialists or distribute them among many reporters, we can suggest that sensor knowledge and expertise should be part of each newsroom’s technical skillset. Good Work Practices for Sensing These are the Tow Center’s observations about how reporters have best worked when using sensors to produce good journalism. Before sensing, articulate your hypothesis. Sensing can be a complex business, especially when journalists are working on an investigative story with high stakes. Two of the environmental journalists who feature in the case studies described very similar processes: They immersed themselves in the scientific literature and very clearly defined hypothesis they were trying to prove. While this seems to be particularly important for projects where journalists are directing the sensing process themselves, in another investigative project, the Sun Sentinel’s story about speeding cops where the journalists were able to access data produced by sensors they did not control, they started with the story they were trying to report and worked to find data to help them. A counterexample may exist, however. The Washington Post’s ShotSpotter report was launched because the reporter found out about a sensor system and imagined what insights it could yield. Still, whether discovering a sensing system before or after a hypothesis is in place, a reporting team should establish the story it’s trying to tell.

On complex stories, work with experts. The previous recommendation concerned journalists researching and defining the phenomena they intend to record. In many cases, journalists may want to go farther by partnering with experts to design and implement their sensing processes. This seems especially important for investigations where journalists are aiming to produce data for comparison with legal standards or health effects. In those cases journalists can expect to have their work examined closely by stakeholders who stand to suffer from the reporting. However, perhaps more importantly, journalists may be producing data that their communities will want to act upon. In that case it is particularly important that journalists’ conclusions are sufficiently accurate and precise. Experts’ experience and skill will provide value throughout the process, from identifying and navigating the legal and ethical considerations during the data collection, to producing correct results. Understand “the Stack.” The journalists featured in these sensing case studies could all talk in detail about how their sensing systems worked. It seems that good journalists’ innate curiosity, combined with the night terror of getting the story wrong, motivated them to understand a lot about every step between the sensed phenomena and the data they collected. As a result, they had more confidence in their story, more ability to communicate it to their audiences, and the security to withstand challenges to their analyses. Specifically, the Public Lab case study suggests that understanding the entire data production process can help journalists hold official interpretations to account. If journalists know that any step in a data production process is problematic, they can duly adjust the credence they assign to the derived data.

Combine sensing with traditional reporting. The value of enriching sensor data with traditional journalistic reporting is twofold. First, if journalists think of sensors as sources, it thereby follows that they should apply the two-source rule. While editors generally prescribe that rule as a protection against factual errors, its benefit also extends to adding detail. Lela Prashad, the satellite sensing expert who contributed an essay to this report’s legal and ethics section, uses the label “ground-truthing” to describe the practice of using extra reporting methods to achieve a more nuanced understanding of the world than remote sensors can collect. The people who live among the sensed phenomena should have a chance to offer their story. Prashad uses the term in the context of information collected from a distance, but the concept seems useful for many different types of sensing. Second, journalists are normally addressing busy, distractible audiences. The story and warmth that comes with interviews and personal observations helps engage audiences, and helps the information journalists report to stick in readers’ minds. Collective Industry Interests These recommendations concern the industry as a whole, and are intended to benefit our communities in general. Journalists have an opportunity and a responsibility to report on sensor systems. Two case studies included in this report feature journalists who used data from sensor systems they discovered in the course of their reporting; the Sun Sentinel accessed highway tollgate data in Florida and The Washington Post reported on data from the ShotSpotter policing system in Washington, D.C. Both examples make it clear that the existence of those systems

and the information they collect were not widely understood. While the Snowden revelations have brought attention to the data produced by email and phone communication, sensor systems have received less focus, even though they extend the potential reach of the security state into more parts of citizens’ lives. This report does not detail those sensor systems that monitor the public for private profit, but other researchers’ work on the potential asymmetry between how public and private interests benefit from big data applies just as well to sensor data as it does to data from other sources.1 Journalists can therefore serve audiences by researching which public and private sensor systems monitor their communities’ lives, not just because those systems are potential sources, but because it is newsworthy to investigate what sensor systems produce, how they are used, and who stands to benefit (or lose) from their data. If journalists and the public start to understand more about the sensor systems that monitor our societies, the next step will be to work through the practical, legal, and ethical questions that should inform how much of the produced data can become public. Advocate for access to data from publicly funded sensor systems. The previous recommendation refers to reporting on sensor systems. This recommendation refers to journalists using data from sensor systems. While acknowledging the privacy considerations detailed in the legal and ethical section of this report, there is no obvious reason that the principles of opengovernment should not apply to data from publicly funded sensor systems. However, at the moment, barriers prevent journalists and the public from accessing data from similar sensor systems in other areas.

In the case of the tollgate data, the Sun Sentinel benefited from Florida’s culture and laws concerning open public records. The reporters were also fortunate that Florida’s highway system is state-owned and operated. Many other states’ (and countries’) highway systems are built and operated within public-private partnerships. We have not tested whether the Sun Sentinel’s story could have been reported in that circumstance, but given that public records laws do not apply to private companies, the odds are bad. As public infrastructure is built through public-private partnerships, an unwelcome side effect may be that data from the embedded sensing systems is not available to all stakeholders. Inspired by The Washington Post’s report, the Tow Center has (along with other journalists) requested data from ShotSpotter installations in a number of other cities. To date, approximately five months after first contact, those cities have refused those public records requests. Police forces have argued against releasing data on the grounds that it could endanger the system itself or that the system is run by a private vendor. Of course, public records requests are often long, contentious processes, regardless of whether the information is produced by sensors or some other source. However, just because delays and refusals have been the historical norm, this does not make them any more defensible for this class of data. Legal and Ethical Recommendations Many of the essays in the legal and ethical section of this report contain recommendations specific to their particular topics. The recommendations below are broadly applicable and high-priority.

Now is the time to play it safe. These are the early stages of using sensors for journalism. That means the laws and ethics are unfamiliar, and journalists are unlikely to be expert in navigating them. Mistakes may hurt the people directly involved in the story, but also prompt legislators to write laws that restrict other journalists from producing good work using similar sensor-based techniques. Review the existing legal and ethical codes. Journalists constantly apply ethical judgments to their work, normally working within the existing industry codes. Nothing in this report suggests that the underlying principles of those codes have become invalid. However, if journalists working with traditional reporting methods apply those codes almost automatically (because they’re expert in doing so), it follows that they would rarely assign much time and attention to that process. Given the relative novelty of sensor techniques, journalists may need to bring the legal and ethical codes of journalism (and of the sciences) to the front of their minds for a period until they become polished in their application in these new settings. Use this report’s essays as a guide for working with your lawyers and ethicists. The authors of this report’s essays have combined their understanding of legal and ethical principles to known and foreseeable uses of sensors in journalism and related fields. However, the essays cannot substitute for a detailed understanding of the specifics of individual reporting projects and the laws and community standards of each jurisdiction journalists may be working within. Therefore, take our authors wisdom as a guide when you seek your own counsel.



Endnotes: The Landscape

1 Indeed computational and data journalism form a strong core of the Tow Center’s research program. Readers of this report may also consider reading the work of Tow Fellows Jonathan Stray, Nicholas Diakopoulos, and Alex Howard.
2 “All the time” is a turn of phrase used in this research study. As those who follow the news about algorithms in financial systems know, sometimes logging data as frequently as once a millisecond is too slow.

Endnotes: Sensors and Sensibilia

1 C.E. is the abbreviation for Common Era, a scholarly alternative naming of the traditional calendar era, Anno Domini (A.D.). Similarly, B.C.E. is an abbreviation for Before Common Era.

Endnotes: The Characteristics of Sensors

1 These qualities are adapted from the widely cited textbook The Sensor Technology Handbook, edited by Jon Wilson. It mainly explores electronic sensors, but the qualities included here are broadly applicable.

Endnotes: Case Studies

1 Readers who would like to study innovative journalistic uses of satellite data should read Paving Paradise, Craig Pittman and Matt Waite’s book about Florida’s vanishing wetlands.

Endnotes: Houston Chronicle

1 The EPA’s health effect screening levels are set at the point equal to the risk of one additional person getting cancer per million of population.

Endnotes: Public Lab

1 The Knight Foundation also funds part of the Tow Center.
2 Emergency Response Management Agency, Deepwater Gulf Response Map, http://gomex.erma. 3 Public Lab, “Wilkinson Bay (Northeast), Louisiana,” 22 Jul. 2010, wilkinson-bay-northeast-louisiana/2010-07-22.

Endnotes: USA Today

1 Peter Eisler and Alison Young, “Some Neighborhoods Dangerously Contaminated by Lead Fallout,” USA Today, 20 Apr. 2012, 04-20/smelting-lead-contamination-soil-testing/54420418/1.
2 Ibid.

Endnotes: Sun Sentinel

1 Mark Puente, “St Petersburg Police Take-home Car Perk Under Scrutiny,” Tampa Bay Times, 24 May 2013, generous-police-perk-on-take-home/2122888.
2 Akintola Akintoye, Matthias Beck, and Cliff Hardcastle, eds., Public-Private Partnerships: Managing Risks and Opportunities (Oxford: Blackwell Science, 2003).

Endnotes: WNYC

1 Guan Yang was looking to make some money for Hack Manhattan, a tax-exempt nonprofit. He’d produced 50 kits with custom circuit boards selling at $20 apiece, across a gross profit of about $8 each. Had he sold all 50, he’d have made Hack Manhattan a profit of about $400. Their more popular breadboard versions had profits of about $4 each.
2 Yes, this can be read as shorthand for “The People Formerly Known As The Audience,” as per media professor Jay Rosen’s seminal blog post of that title from June 30, 2006. See

Endnotes: Planet Money

1 The Jensen brothers won’t disclose the insurance company with whom they work, saying only that theirs was a difficult policy to procure. At least two companies, Hill & Usher and Transport Risk, do write insurance policies for camera drones working inside the United States.
2 Duy Linh Tu, “Video Now: The Form, Cost, and Effect of Video Journalism,” Tow Center for Digital Journalism, Columbia University, Feb. 2013,

Endnotes: The Washington Post

1 This is an approximate figure. In fact, the accuracy of latitude and longitude numbers with decimal points (or decimal degrees) changes depending on the location’s proximity to the equator. The farther away from the equator, the more precise the number.

Endnotes: Industry Overview

1 The FAA also sends letters to small drone pilots who fly their craft “for benefit,” warning that the next action may be prosecution if the pilots don’t stop flying.
2 The NPPA’s general counsel, Mickey Osterreicher, led the project. The Tow Center helped to design the survey with the aid of Matt Waite, a professor of journalism practice at the University of Nebraska; and Charles D. Tobin and Charles R. Naftalin, both partners at the law firm of Holland & Knight.
3 That figure may have double-counted some drone use, where multiple staff members of a single company referred to the same drone flights. It may have also missed uses by companies whose employees did not see the survey. The survey does not provide data that can be extrapolated to the entire industry.
4 The Daily News posted it to its servers and incorporated the video and stills into its news coverage. Wilson was flying a DJI Phantom, a threepound quad copter that retailed for less than $1,000.
4 Bill Hutchinson, “Drone Captures Scene at East Harlem Explosion That Flattened Two Buildings,” New York Daily News, 13 Mar. 2014, harlem-explosion-scene-video-article-1.1719988.
5 Beto Lopez, “Valentine’s Day Pillow Fight 2014,”
6 Brian Emfinger, “Arkansas Tornado Damage Aerial Video 4-27-2014,” YouTube, 27 Apr. 2014,
7 Duy Linh Tu, “Video Now: The Form, Cost, and Effect of Video Journalism,” Tow Center for Digital Journalism, Columbia University, Feb. 2014,
8 Jeremy Barr, “Photographer Says Spokesman-Review’s Drone-shot Video Occurred in a Gray Area,’ ” Poynter, 3 Jan. 2014, reviews-drone-shot-video-occurred-in-a-gray-area/.

Endnotes: Syed

1 Louis Brandeis and Samuel Warren,“The Right to Privacy,” 4 Harv. L. Rev. 5 (1890).
2 See Restatement (Second) of Torts Section 652.
3 See, e.g,. ACLU v. Alvarez (7th Cir. 2012); Glik v. Cunniffe (1st Cir. 2011). 4 Askins v. U.S. Dept. of Homeland Sec. (S.D. Cal. Sept. 20, 2013).
5 Of course, Jones dealt with the law enforcement use of drones under a Fourth Amendment standard. This would not apply to private newsgathering entities. Nevertheless, the court’s logic shows that common-law trespass doctrine is alive and well, and that courts are not afraid to use it when dealing with sensing devices.
6 The doctrine of nuisance, however, would address any invasion that causes adverse, substantial, or unreasonable interference with the enjoyment of one’s own land, even without the physical invasion required by trespass. A sensor-equipped drone or a sensor emitting disruptive lights or frequencies could be subject to this claim, so long as there is demonstrable harm.
7 Whether one can affix sensors on public property would involve a case-by-case analysis. Some jurisdictions have restrictions on what can be placed on trees in a public park, for example.
8 Audio recording of conversations is well-regulated by both federal and state law. Most states permit audio recording so long as one party, which could be the recording party, to the conversation consents. A handful of states—California, Connecticut, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Pennsylvania, and Washington—prohibit audio recording of a conversation unless all parties to that conversation have consented. Of course, this requirement of consent extends only to conversations where there is some expectation of privacy, but any sensor that can collect audio data should be generally aware of these state-level requirements.
9 Public officials, public figures, and limited-purpose public figures must demonstrate a higher standard of fault by the reporter. These individuals would have to show that the reporter either knew that the reporting (such as the sensor readings or analysis) was false, or entertained serious doubts as to its veracity. This turns on the reporter’s actual state of mind at the time of publication. This is a very high burden of proof. 10 Restatement (Second) of Torts, Sec. 652E.

Endnotes: Culver

1 Daily Drone: Seemingly Endless Waters in Mississippi, YouTube, 12 May 2011, com/watch?v=MI2d3ui2vYw.
2 “Cicada Tracker,” Radiolab and WNYC,
3 Andy Stettler, “How We Crowdsourced Breaking News: Bomb Device’ Reported at King of Prussia Mall,” The Reporter, 4 Jan. 2012, crowdsourced-breaking-news.html.
4 Alison Young and Peter Eisler, et al., “Ghost Factories,” USA Today, 19 Apr. 2012, http://
5 Sally Kestin and John Maines, “Above the Law,” Sun Sentinel, speeding-cops/.
6 C.G. Christians, “The Ethics of Privacy,” Journalism Ethics: A Philosophical Approach, Ed. Christopher Meyers (Oxford: Oxford University Press, 2010), 203.
7 Christians (2010), 205.
8 Daniel J. Solove, Nothing to Hide: The False Tradeoff Between Privacy and Security (New Haven: Yale University Press, 2011).
9 Yves-Alexandre de Montijoye, et al., “Unique in the Crowd: The Privacy Bounds of Human Mobility,” Scientific Reports, 25 Mar. 2013, srep01376.html.
10 Jim Colgan, “Mapping the Storm Clean-up,” WNYC, 30 Dec. 2010, 2-mapping-storm-clean/.
11 Solove (2011), 30.
12 Limor Fried, “Minimizing Risk is Easy: Adopt a Bill of Rights,” New York Times, 8 Sep. 2013, for-the-internet-of-things.

Endnotes: Stearns

1 “Poynter Publishes Definitive New Journalism Ethics Book,” Poynter Institute, 11 Sep. 2013, book.
2 C.W. Anderson, “The Public is Still a Problem, and Other Lessons From Ŕebuilding the News,’” Nieman Journalism Lab, 18 Jan. 2013, still-a-problem-and-other-lessons-from-rebuilding-the-news/.
3 “The Laws, Ethics, and Politics of Sensor Journalism,” panel at the Sensor Journalism Workshop, Tow Center for Digital Journalism, Columbia University, 6 Jun. 2013, sensors-the-laws-ethics-and-politics-of-sensor-journalism.
4 Fergus Bell and Eric Carvin, “Social Newsgathering: Charting an Ethical Course,” Online News Association, 6 Mar. 2014, course/.
5 Poynter replaced the traditional “minimize harm” philosophy with “engage community as an end rather than as a means.” See a discussion of the changes here: the-next-journalism/223657/why-be-transparent-is-now-a-better-ethical-principle-than-actindependently/.
6 Eloïse Gratton, “If Personal Information is Privacy’s Gatekeeper, then Risk of Harm is the Key: A Proposed Method for Determining What Counts as Personal Information,” Albany Law Journal of Science and Technology, Vol. 24, No. 1 (2013),
7 Ibid.
8 Alistair Croll, “Big Data is Our Generation’s Civil Rights Issue, and We Don’t Know It,” Radar, 2 Aug. 2012, we-dont-know-it.html.
9 “Civil Rights Principles for the Era of Big Data,” The Leadership Conference, 2014, www. 10 For a good discussion of some of these issues related to the history of surveillance and communities of color, see the summary of a panel discussion held by Free Press, www.freepress. net/blog/2013/11/01/panelists-say-surveillance-nothing-new-communities-color and remarks from Amalia Deloney at the Media Consortium meeting, delivered in February 2014, www.magnet. org/sites/default/files/amaliaTMCcommentsmediaconsortium.pdf.
11 Jonathan Peters, “Updating the Privacy Protection Act for the Digital Era: Law Protecting Journalists From Searches Didn’t Anticipate Cloud Computing,” Columbia Journalism Review, 30 Jan. 2012,
12 Alistair Croll, 2012.

Endnotes: Schroyer

1 Irving Washington, “Be Open to Open Source,” Online News Association, 30 Jan. 2013,
2 David Kline and Kevin Rivette, “Discovering New Value in Intellectual Property,” Harvard Business Review, Jan./Feb. 2000. R. Polk Wagner, “Information Wants to Be Free: Intellectual Property and the Mythologies of Control,” Columbia Law Review, Vol. 103, No. 4 (May 2003) 995–1034.
3 Brian Heater, “Nest Protect Review: A Smoke Detector for the Smartphone Generation,” Engadget, 20 Dec. 2013,
4 Nicholas Collins, Handmade Electronic Music: The Art of Hardware Hacking (New York: Routledge, 2006).
5 Nicholas Diakopoulus, “Algorithlmic Accountability Reporting on the Investigation of Black Boxes,” Columbia Journalism School, Tow Center for Digital Journalism, Dec. 2013,
6 Rakesh Sharma, “Hardware Is Not Just Śoftware Wrapped In Plastic,’ According to VC Bilal Zuberi,” Forbes, 20 Oct. 2013, software-wrapped-in-plastic-according-to-vc-bilal-zuberi/.
7 James Bessen, Jennifer Ford, and Michael Meurer, “The Private and Social Costs of Patent Trolls,” Regulation, Winter 2011–2012, regulation/2012/5/v34n4-1.pdf.
8 Alexander Hars and Shaosong Ou, “Working for Free? Motivations for Participating in Open- Source Projects,” International Journal of Electronic Commerce, Vol. 6, No. 3 (Spring 2002) 25–39.
9 Matt Wynn, “Introduction to Open-source GIS Tools for Journalists,” Poynter Institute, 19 Sep. 2011, for-journalists/.
10 Sukwon Choi, et al., “Micro Sensor Node for Air Pollutant Monitoring: Hardware and Software Issues,” Sensors, 9.10 (2009): 7970–7987.
11 D. M. Holstius, et al., “Field Calibrations of a Low-cost Aerosol Sensor at a Regulatory Monitoring Site in California,” Atmospheric Measurement Techniques Discussions, 7.1 (2014) 605–632.
12 Hars and Ou, 2002.
13 Chris Anderson, Free: The Future of a Radical Price (New York: Hyperion, 2009).
14 Matthew Schroyer, “A Call for Journalists and Makers to Join Hands Around IOT and Evidencebased Journalism,” The Mental Munition Factory, 27 Feb. 2014, www.mentalmunition. com/2014/02/a-call-for-journalists-and-makers-to.html.
15 Karim R. Lakhani and Eric von Hippel, “How Open Source Software Works: Free’ User-to-user Assistance,” Research Policy, Vol. 32, Iss. 6 (June 2003), 923–943.
16 Josh Lerner and Jean Tirole, “Some Simple Economics of Open Source,” The Journal of Industrial Economics, Vol. 50, No. 2 (Jun. 2002), 197–234. See also Margaret Elliot and Walt Scacchi, “Communicating and Mitigating Conflict in Open Source Software Development Projects,” 2002, melliott/commossd.pdf.
17 Ethan Mollik, “The Dynamics of Crowdfunding: An Exploratory Study,” Journal of Business Venturing, Vol. 29, Iss. 1 (Jan. 2014), 1–16.
18 Joel West, “How Open is Open Enough?: Melding Proprietary and Open Source Platform Strategies,” Research Policy, Vol. 32, Iss. 7 (Jul. 2003), 1259–1285.
19 Ibid.

Endnotes: Hord

1 FCC Office of Engineering and Technology Web site,
2 FCC Enforcement Bureau Web site,
3 Code of Federal Regulations (CFR)-2010, Title 47, Vol. 1, Part 15, Subpart A, section 3k.
4 CFR-2010, Title 47, Vol. 1, Part 15, Subpart B, section 101a.
5 CFR-2010, Title 47, Vol. 1, Part 15, Subpart C, section 201a.
6 CFR-2010, Title 47, Vol. 1, Part 15, Subpart A, section 23a.
7 The last major revision to the FCC rules was in 1989. Kenneth R. Carter, “Unlicensed to Kill: A Brief History of the Part 15 Rules,” info, Vol. 11, Iss. 5 (2009) 8–18.
8 Federal Communications Commission, Order and consent decree, Part ii, 31 Apr. 2012,
9 CFR-2010 Title 47, Vol. 1, Part 15, Subpart A, section 3p.
10 Federal Communications Commission, Memorandum of opinion on the above cited case, 30 Apr. 2008, 11 See, for instance, the “Wave Bubble” cellular jamming device, wavebubble/index.html.
12 For a good example of trouble caused by using illegally marketed electronics, see http://, which is an enforcement citation regarding a wireless video camera system.
13 CFR-2010, Title 47, Vol. 1, Part 15, Subpart C, section 212.

Endnotes: Cooper

1 Diana Marina Cooper, “A Licensing Approach to Regulation of Open Robotics,” Robot Law, eds. Ryan Calo, Ian Kerr, and Michael Froomkin (Edward Edgar, forthcoming 2015), initially presented at the We Robot Conference, Stanford Law School, Apr. 2013, werobot/files/2013/04/CooperDiana.pdf. Ryan Calo, “Open Robotics,” Maryland Law Review, Vol. 70, No. 3 (2011), 103.
2 Ibid.
3 “What is Open Journalism and What is Its Appeal?” Open Journalism Report of the World Association of Newspapers and News Publishers, 2012, open-journalism-executive-summary-download.
4 Ibid.
5 Diana Marina Cooper, 2013.
6 Ibid.
7 Ryan Calo, 2011.
8 Ibid.
9 Ibid.
10 Ibid.
11 Diana Marina Cooper, 2013.
12 Jason Koebler, “Oregon Company to Sell Drone Defense Technology to Public,” US News & World Reports, 15 Mar. 2013, defense-technology-to-public.
13 Diana Marina Cooper, 2013.
14 Diana Marina Cooper, “Effects of Cloud Computing on Open Source Compliance,” Linux Journal, Iss. 229, May 2013, compliance.
15 Mozilla Foundation, About MPL 2.0: Revision Process and Changes FAQ, MPL/2.0/Revision-FAQ.html.
16 American Society for Photogrammetry and Remote Sensing, “Code of Ethics,” Remote-Sensing.html.
17 Ibid.

Endnotes: Waite and Osterreicher

1 Federal Aviation Administration, “Busting Myths About the FAA and Unmanned Aircraft,” 26 Feb. 2014,
2 Federal Aviation Administration, The FAA Modernization and Reform Act of 2012, 1 Feb. 2012,
3 Ben Gielow, “Small UAS Rule Release Delayed Until November 2014,” Association for Unmanned Vehicle Systems International, 14 Jan. 2014,
4 Huerta v. Pirker, 2014,
5 Federal Aviation Administration, “FAA Statement,” 7 Mar. 2014,
6 Federal Aviation Administration, Integration of Civil Unmanned Aircraft Systems (UAS) in the National Airspace System (NAS) Roadmap; Notice of Final Privacy Requirements for UAS Test Sites; and UAS Comprehensive Plan, Nov. 2013, cfm?newsId=15334.
7 Federal Aviation Administration, Integration of Civil Unmanned Aircraft Systems (UAS) in the National Airspace System (NAS) Roadmap, First Edition (2013), 52, initiatives/uas/media/UASRoadmap2013.pdf.
8 Ibid., 35.
9 Ibid., 58.
10 Federal Aviation Administration, UAS Comprehensive Plan, 6 Nov. 2013, officeorg/headquartersoffices/agi/reports/media/UASComprehensivePlan.pdf.
11 Ibid., 33.
12 U.S. House, Explanatory Statement to the Consolidated Appropriations Act, 2014,
13 Ibid., 6.
14 Ibid., 7.
15 Texas House, H.B. No. 912, State of Texas, 17 May 2013,
16 Ibid.
17 Idaho Senate, S.B. No. 1134, State of Idaho, 2013,
18 Ibid.
19 Iowa House, H.B. 2289, State of Iowa, Apr. 2014,
20 Ibid.
21 Ibid.
22 Ibid.
23 Georgia House, H.B. No. 846, State of Georgia, 28 Jan. 2014,
24 Washington House, H.B. No. 2178, State of Washington,
25 Ibid.
26 Washington State Legislature, RCW 9.92.020,
27 H.B. No. 2178.
28 Ibid.
29 Ibid.
30 Wisconsin State Legislature, A.B. 203, 2013,
31 Ibid.
32 Ibid.

Endnotes: Sullivan

1 An aurei was a gold coin and monetary unit of ancient Rome, used from Caesar to Constantine I.
2 Associated Press, “The City; Jets Sued on Death as Result of Show,” 18 Nov. 1981, retrieved from
3 A person injured by a sensor, or drone deploying a sensor, might also bring a claim against the manufacture of the drone or sensor, most likely alleging strict liability, which holds the manufacturer liable if the product was defective and a consumer of the product was injured. A negligence claim, however, is far more likely.
4 Each state has its own rules, but as a general matter state courts rely heavily on the Second Restatement of Torts, a treatise that surveys and summarizes the current state of the law of torts in the country.
5 Different jurisdictions have established certain exceptions that limit duty within certain categories of public policy concerns (including the social value and the usefulness of the conduct), of which newsgathering could theoretically be one. No case has yet tested this proposition.
6 This calculus of negligence, called the Hand formula, has limited applicability.
7 Contributory and comparative negligence are important but complex doctrines and have been set aside here in favor of a more basic overview. They would likely arise in a case with a varied set of facts not contemplated by our hypothetical.

Endnotes: Graves

1 Bill Kovach and Tom Rosenstiel, The Elements of Journalism (New York: Three Rivers Press, 2001) 41.
2 Barry Sussman, “Watergate, 25 Years Later: Myths and Collusion,” Making News: A Study in the Construction of Reality, auth. Gaye Tuchman (New York: The Free Press, 1997), available at
3 Dean Starkman, “The Hamster Wheel,” Columbia Journalism Review, 14 Sep. 2010, coverstory/thehamsterwheel.php?page=all. See also Pew Research Center, “The Changing Newsroom,” 21 Jul. 2008,
4 Sylvain Parasie, “Data-driven Revelations? Continuities and Changes in the Epistemology of Investigative Journalism,” Digital Journalism (forthcoming).
5 Philip Meyer, Precision Journalism: A Reporter’s Introduction to Social Science Methods (Indiana University Press, 1973), 3. See also Sylvain Parasie and Eric Dagiral, “Data-driven Journalism and the Public Good: Ćomputer-assisted-reporters’ and programmer-journalistsín Chicago,” New Media & Society, 18 Nov. 2012.
6 Chris Anderson, “The End of Theory: The Data Deluge Makes the Scientific Method Obsolete,” Wired, 23 Jun. 2008.
7 Danah Boyd and Kate Crawford, “Critical Questions for Big Data,” Information, Communication & Society, Jun. 2012.
8 Michael Fienen, “We Speak for the Data,” Ground Water, Feb. 2013.
9 Katherine Fink and Michael Schudson, “The Rise of Contextual Journalism, 1950s−2000s,” Journalism, Feb. 2013. See also Kevin Barnhurst, “The Makers of Meaning: National Public Radio and the New Long Journalism, 1980–2000,” Political Communication, 2003.

Endnotes: Stauffer

1 J.G. Proakis and D.K. Manolakis, Digital Signal Processing: Principles, Algorithms and Applications, 4th Edition (Upper Saddle River, N.J.: Prentice Hall, 2006).
2 FlowingData, “Where People Run in Major Cities,” 5 Feb. 2014,
3 Margaret McKenna, “Beautiful Maps, and the Lies They Tell,” CodeKeeper,
4 U.S. Environmental Protection Agency, “Guidance on Choosing a Sampling Design for Environmental Data Collection,” EPA QA/G-5S, Washington, D.C., 2002.
5 Bonnie Berkowitz, et. al., “Shots Heard Around the District,” Washington Post, 2 Nov. 2013,
6 L.A. Currie, “Detection and Quantification Limits: Origins and Historical Overview,” Analytica Chimica Acta, 391, 1998, 127–134.
7 U.S. Environmental Protection Agency, 2002.
8 National Oceanography and Atmospheric Administration, “Accuracy Versus Precision,” NOAA Celebrates 20 Years, 2007.
9 Ohio Environmental Protection Agency, “The 3 Levels of Credible Data,”
10 Hach Method 10360, “Luminescence Measurement of Dissolved Oxygen in Water and Wastewater and for Use in the Determination of BOD5 and cBOD5,” 2011,
11 Margaret McKenna, “Beautiful Maps, and the Lies They Tell.”
12 Jeff Shaner, “Smartphones, Tablets and GPS Accuracy,” ArcGIS Resources, 15 Jul. 2013,
13 Sally Kestin and John Maines, “Cops Among Florida’s Worst Speeders, Sun Sentinel Investigation Finds,” Sun Sentinel, 11 Feb. 2012, fl-speeding-cops-20120211,0,3706919.story.
14 National Information Standards Organization, Understanding Metadata (Bethesda: NISO Press, 2004).
15 A. G. Wiggins, R.D. Newman, and K. Crowston Stevenson, “Mechanisms for Data Quality and Validation in Citizen Science,” Proceedings of IEEE eScience Conference, Stockholm, Sweden, 2011.
16 C.C. Wilderman and J. J. Monismith, “Marcellus Shale Gas Extraction: A Study Design and Protocol for Volunteer Monitoring,” 2012, files/2012/07/Marcellus-Shale-Volunteer-Monitoring-Manual-2.0.pdf.

Endnotes: Recommendations

1 Danah Boyd and Kate Crawford, “Six Provocations for Big Data,” A Decade in Internet Time: Symposium on the Dynamics of the Internet and Society, 21 Sep. 2011, or


Author Biographies


Fergus Pitt Fergus Pitt is a research fellow at the Tow Center for Digital Journalism. He worked for the Australian Broadcasting Corporation for twelve years, where he started producing and reporting radio current affairs, stepped sideways into online development and production, then ran projects, wrote strategy and managed product development. He received his M.A. in Journalism (Business & Economics) from Columbia University in New York, and a B.A. in Journalism from RMIT University in Melbourne, Australia. He tweets from @fergle.

Charles Berret Charles Berret is a Ph.D. candidate at the Graduate School of Journalism at Columbia University. He researches the cultural and technological history of typographic media, especially the development of digital typography from earlier platforms. More broadly, he is interested in visual culture, semiotics, machine aesthetics, and the interconnected histories of technology, graphic design, and the book.

Before arriving at Columbia, Charles worked as a journalist in Cairo and a book reviewer for literary journals. He studied philosophy at the University of Michigan and journalism at Northwestern University. He tweets from @cberret.

Nabiha Syed Nabiha Syed is an Associate in the Emerging Technologies section of Levine Sullivan Koch & Schulz law firm and a visiting fellow at Information Society Project Fellow at Yale Law School. She was formerly the First Amendment Fellow at The New York Times and is a Co-Founder of the Media Freedom and Information Access legal clinic and Drone U. Her legal practice includes matters of newsgathering including drones and sensors, litigated access to courts, and freedom-of-information issues. She also writes on legal, political, and social issues concerning drone technology. Her JD is from Yale Law School. She holds a Master of Studies in Legal Research, Comparative Media Law and Policy from Oxford University, and a B.A. from Johns Hopkins University. She tweets from @nabihasyed. Kathleen Culver Kathleen Culver is an assistant professor in the University of Wisconcin- Madison School of Journalism & Mass Communication and associate director of the Center for Journalism Ethics. She also serves as visiting faculty for the Poynter Institute for Media Studies and curates education content for PBS Media Shift.

She focuses on the ethical dimensions of social tools, technological advances and networked information. She combines these interests with a background in law and the effects of boundary-free communication on free expression. She holds a B.A., Journalism and Mass Communication (Reporting), an M.A., Journalism and Mass Communication (Media Law) and a Ph.D., Mass Communication (Media Law) all completed at the University of Wisconsin-Madison. She tweets from @kbculver.

Josh Stearns Josh Stearns is the Director, Journalism & Sustainability for the Geraldine R. Dodge Foundation. Formerly he was the Press Freedom Director at Free Press. His work focuses on building community around the news through organizing, creative storytelling, high impact events, public education and collaborative problem-solving. He currently manages three national campaigns focused on strengthening press freedom, expanding public media and opposing media consolidation. He is a founding board member of the Freedom of the Press Foundation. His graduate degree in American Studies is from the University of Massachusetts, Amherst. He tweets from @jcstearns.

Matthew Schroyer Matthew Schroyer is a communications specialist, social network analyst, and data journalist based in Urbana, Illinois. He develops unmanned aircraft systems (commonly called drones) and sensor node technology for journalism and STEM education. He is the founder and president of the Professional Society of Drone Journalists. He holds a master’s degree in journalism from the University of Illinois Urbana-Champaign. He tweets from @matthewryan.

Mike Hord Mike Hord is an Electrical Engineer at SparkFun Electronics, designing electronic parts for makers. He has been in the field for ten years, six of which have involved EMC regulations for the EU or the US. When not hacking electronics, he is hacking the future by raising two young children. He holds a Bachelor of Science in Electrical Engineering, from North Dakota State University. He tweets from @uptownmaker.

Diana Marina Cooper Diana Cooper is the Head of Unmanned Aerial Systems and Robotics, at the LaBarge Weinstein LLP law firm in Ottawa, Canada. She also specializes in intellectual property, technology and privacy. In her spare time, Diana builds hobby robots and flies her own drone. Diana’s interest in robots and drones led her to write a number of academic papers focused on the regulation of robotics. She graduated from University of Ottawa with a JD, Cum Laude, Law & Technology, and obtained a Bachelor of Arts from Ryerson University and a Masters of Arts from McMaster University. She was also awarded the Osler, Hoskin & Harcourt LLP Prize in Business Law. She tweets from @DianaMCooper.

Matt Waite Matt Waite is a Journalism Professor at the University of Nebraska-Lincoln, where he teaches digital journalism, programming and data visualization and founded the Drone Journalism Lab. In 2009, he won a Pulitzer Prize for developing the fact checking site while working as a News Technologist at the St. Petersburg Times. Previously, he worked as a staff writer and investigative reporter at the same newspaper. He holds a B.A. in journalism from the University of Nebraska-Lincoln, and at time of writing is taking the FAA’s ground school classes to get The Drone Journalism Lab a Certificate of Authorization. He tweets from @mattwaite.

Mickey Osterreicher Mickey H. Osterreicher is the General Counsel for the National Press Photographers’ Association. He is experienced in contract, media, copyright and First Amendment Law and has been actively involved in matters such as cameras in the courtroom, the federal shield law, media access, public photography, orphan works and copyright infringement. He worked as a photojournalist with material appearing in The New York Times, Newsweek and USA Today as well as on ABC World News Tonight, Nightline, Good Morning America, NBC Nightly News and ESPN. He went to law school at the State University of New York at Buffalo and holds a B.S. in Photojournalism. He tweets from @nppalawyer.

Deirdre Sullivan Deirdre Sullivan practices in the legal department of The New York Times Company. Deirdre’s practice focuses on the business, legal and policy aspects of new media issues that arise across digital platforms, including digital ad operations and related privacy concerns. She counsels all of the The New York Times Company brands. Prior to joining The New York Times Company, Deirdre was an associate at Fried, Frank, Harris, Shriver & Jacobson LLP. Deirdre is a graduate of the Northwestern University School of Law. The views she expresses in this report are her own and do not reflect the views of The New York Times Company.

Lucas Graves Lucas Graves is an Assistant Professor in the School of Journalism and Mass Communication at the University of Wisconsin-Madison. His research focuses on new organizations and practices in the emerging news ecosystem, and more broadly on the challenges that digital, networked communications pose to established media and political institutions. He is currently at work on a book about the fact-checking movement in American journalism. He co-authored the first two reports published by the Tow Center, on internet advertising metrics and on business models for online news. He worked as a magazine journalist with work appearing in Wired magazine, Interview, Time Out New York, Food & Wine, The Brooklyn Rail, and many other publications. Graves received his doctorate and an M.S. in journalism from Columbia University. He holds a B.A. in political science from the University of Chicago. He tweets from @gravesmatter.

Beth A. Stauffer Beth Stauffer is an AAAS Science & Technology Policy Fellow working in the Environmental Protection Agency’s Innovation Team. Her research interests are grounded in phytoplankton ecology and water quality in marine and freshwater ecosystems. She is also intensely interested in the development and application of new tools to study and model aquatic ecosystems over time and space. Throughout her graduate career, she was involved in the Center for Embedded Networked Sensing (CENS) exploring technologies for embedded, autonomous sensing and novel technologies for aquatic research. She was awarded her Ph.D. from the University of Southern California and held the Lamont-Doherty Earth Observatory postdoctoral fellowship at Columbia University. Her B.S. in Marine Science is from the University of Miami.

Lela Prashad Lela Prashad is the Chief Data Scientist and Co-Founder of NiJeL, a company which helps people and organizations tell their stories and become more efficient and effective through data and maps. Previously, she directed the 100 Cities Project at Arizona State University and is still active in urban research with ASU. She established and ran the water program for the environmental non-profit Arizona PIRG and worked with the United States Geological Survey’s Earthquake Hazards Team, and Arizona Departments of Environmental Quality and Water Resources. Lela holds an Master of Science in Geological Sciences from ASU and a Bachelor of Science in Geosciences from Trinity University in San Antonio, Texas. She tweets from @lelap.