Ethics, Laws, Sensors and Journalism — A Starting Point
We’re publishing this blog post as a year’s end marker for our sensor journalism research. At the very start of June, when we held our two-day workshop on sensors and journalism, many participants called for the Tow Center to develop legal and ethical resources for people entering the field. Over the last six months, while The Tow Center’s research into sensors and journalism has progressed, through developing projects and consulting experts, we’ve made good headway on that front. So the article below serves a number of purposes; we’re exposing our current thinking about laws and ethics; the story so far. It’s a step towards publishing a definitive, reliable resource for journalists, managers, policy analysts and lawyers, which we will be doing in the first half of 2014.
We’re distributing this article now to host and frame a discussion, which we’d like you to join. Please, read this post and tell us whether you think the groupings make sense, whether we’ve missed important considerations, if there are bodies of research that should be brought into play.
This post has been written in the USA, by an Australian working closely with a Canadian, directed by an Englishwoman. Unsurprisingly, it draws on concerns and knowledge of those countries’ journalism industries — so your responses that pull in ideas and views from other countries would be particularly welcome.
We’ll monitor and respond to comments on this page, or if you’d prefer to contribute in private you can email ferguspitt@columbia.edu
The phrase ‘Sensor Journalism’ doesn’t really have much meaning. But it’s a useful contrivance, which acknowledges that sensors — if we’re using them as reporting tools — need to be handled with distinct skills and knowledge to produce good data for journalism.
This article offers a survey of the legal and ethical considerations that The Tow Center believes are relevant for journalists who are using sensors as reporting tools. From our research thus far, we believe that the considerations fall into four areas;
- Privacy & Surveillance
- Accuracy, Interpretation & Communication
- Building & Acquisition
- Deployment & Operation
We’ve noticed that the first two groups, “Privacy & Surveillance”, “Accuracy, Interpretation & Communication” have an intellectual feel: journalists may need to pay particular attention to their thought processes and quite complex ethical considerations about how their work interacts with other parts of society. I’ve expanded on these below, but there is much more work to be done in this realm.
The second two groups; “Building & Acquisition” and “Deployment and Operation” have a practical focus, centering on objects that exist in physical space (although they may accelerate the integration of cyberspace and physical space). This physicality introduces gnarly legal problems — not insoluble, but certainly potent.
Of course, these groupings make somewhat artificial distinctions; for example journalists who are in a sensor-building phase will probably need to consider surveillance ethics and laws simultaneously with software licensing concerns.
Although I wrote this article with an underlying assumption that some journalists will be deploying sensors themselves, the legal and ethical issues apply just as well if the sensors are being built and administered by a third party, such as a university or an environmental protection body (although in those cases the responsibilities may shift, depending on the relationship between the journalist and the party running the sensor).
Indeed, examples of journalism that incorporate sensor data include projects that operated on a range of different models. When USA Today reported The Smokestack Effect, which monitored air quality around schools, they deployed sensors “under guidance of scientists from Johns Hopkins University and the University of Maryland School of Public Health”. The WNYC Cicada Tracker team built soil temperature sensor units to their own design, (which was subsequently improved by their community). The Washington Post’s ShotSpotter investigation worked with data from microphones installed by a private company, for the primary benefit of the Washington DC Police Department.
Later publications by The Tow Center will explore the landscape of sensors in depth, but for journalists, lawyers and ethicists who may only be starting to think about the field, we offer these sensor continuums so that readers start with an reasonable mental model:
- Autonomous ←→ Controlled: A handheld radiation monitor is an example of a very controlled sensor. A remotely controlled drone (UAV) is more autonomous and an air-quality sensor installed then left to run is even further along the autonomy spectrum.
- Near ←→ Far: A satellite rotating the earth is far, a fitbit activity monitor is near.
- Official ←→ Ad Hoc: The EPA’s network of air quality sensors are quite official, the US Embassy’s Air Quality monitor in Beijing is probably somewhere near the middle, whereas kites carrying air quality sensors over Beijing are quite ad-hoc.
- High Resolution ←→ Low Resolution: An accelerometer recording 100 readings a second has high temporal resolution, whereas a camera set up to record 1 pixel for 50 meters of land has low spatial resolution.
As a substitute for a sensor typology, readers who want to get more familiar with the variety of sensors might review the sensor catalogue at mouser to get a good feel for what can be sensed (although their catalogue is not exhaustive; it doesn’t have any water contaminant sensors).
For each different type of sensor system, operating at different parts of the continuums above and producing data on different journalistic subjects, the legal and ethical issues below will apply in varying ways. They may even attract extra legal and ethical considerations that I haven’t noted here. This is an early outline of the issues, and we’ll be expanding and deepening it over the coming months, with specific legal references, lines of research and tools for journalists to navigate their way.
Group 1: Privacy And Surveillance
When journalists or researchers use sensors to collect data that records people and their traces, intentionally or not, they should probably review their legal and ethical considerations through a privacy and surveillance lens. Journalists’ ethical guidelines and various bodies of law already protect privacy (although the protections vary between jurisdictions, the location of the subject and the type of information being recorded — as do community expectations). Likewise, the Snowden revelations have sparked a debate about the methods and extent of state-run surveillance, even as the awareness and concerns about data mining spread beyond communities of the technologically sophisticated, academics and privacy advocates. The debate comes after decades of technical innovation that has increased data’s utility in commercial settings, especially in marketing. Journalists may want to research the developments in deanonymization and whether the data they are collecting and storing can be used in new ways. That might be direct use, or providing a missing mosaic piece for other available data sets.
We’re grouping the following considerations and questions within the broad area of privacy and surveillance:
Consent & Notification
While the concept of consent is clearly relevant for sensing, it is harder to understand how and when to obtain consent and whether opt-ins or opt-outs are appropriate. In American cities, we are used to seeing signs informing us that we’re being recorded on camera. Retailers such as Nordstrom that have tracked cell-phones through their stores have also installed customer-facing signage about the technology. However, news crews record crowd footage in public places without obtaining consent, and Google’s street view cars were not banned from collecting WiFi network IDs without consent (although they were required to destroy “payload data“).
As noted above, those decisions may vary with the location of the subject, the location of the sensor, the institution or person in charge of the sensor, the type of data/content being collected.
A lot of the laws and ethical guidelines in this realm hinge on the concept of reasonable expectations — but in a rapidly changing environment with a fast-moving debate, attitudes may be moving faster than legal tests.
Defamation & Libel
When recording and distributing personally identifiable data, journalists risk defaming people. Even in jurisdictions where truth is a defense, sensors may be accurate insofar as the physical thing they record did happen, but they may not collect the full context and therefore depict a person in such a way that brings them into “disrepute among right-thinking people”. Some jurisdictions call this concept ‘false-light’.
It seems that — as with many types of data journalism — when journalists are reporting with sensors, they’re likely to be interpreting data; drawing inferences and conclusions. There may be greater risks of inaccurate interpretations or over-stating certainty, especially when employing unfamiliar data collection techniques. I’ve included more detail on optimizing data collection for accuracy below. As noted in the privacy and surveillance paragraphs above, if the sensor data is being used to identify people (using for example, using facial recognition, or the ID of a smartphone) journalists may want to take particular care that their data is accurate.
Vulnerable Populations
Some ethical codes require that journalists adjust their approach to privacy when deliberately or potentially recording particular vulnerable populations: young people, people who are permanently or temporarily mentally impaired, people who are involuntarily present in a space, or people who are involuntarily participating in an activity (prisoners, for example). When sensors are deployed autonomously, where journalists are not always observing the people in the space, the risk of inadvertently recording vulnerable populations may increase.
However, it appears that while much of the existing literature on young peoples online behavior and attitudes reveals that some teenagers have adopted sophisticated techniques to maintain privacy, the research I’ve encountered only extends to an environment when the internet is on computers and phones, as opposed to embedded in many objects which are capable of sensing. The “internet of things” may pervade more spaces where vulnerable populations congregate or pass by.
Ownership, Access and Control Over Sensor Data
In many jurisdictions, individuals have the right to know and/or correct information ‘about’ them that is held by some other parties. Journalists may want to research whether those kinds of laws apply to their company and the kind of data their sensors are creating. The movement towards data transparency has even reached Axiom, one of the larger marketing data miners in the USA who have provided a portal for consumers to check the data the company holds about them (although some critics note that the data they expose to the public is a subset of what’s available to paying clients, and some of the data is wrong).
In recent ‘Internet of Things‘ coverage in the New York Times, Limor Fried, the founder of DIY electronics distributor Adafruit, advocated the concept of an IoT Bill of Rights. It included the right for public examination of the designs and the data collected by devices, and the to refuse to be recorded, amongst other things.
On the face of it, there is a tension between the idea that data should be accessible and controllable by its subject and journalist’s custom of keeping data (or observations, or intelligence) confidential until publication.
Becoming an arm of surveillance states
If a journalist is collecting data through sensors or any other means, they may need to consider whether they are unintentionally extending the reach of surveillance agencies. While journalistic culture and ethics go against revealing sources and turning over notebooks, the ability of news organizations to secure their data from eavesdropping and hacking varies with their training, discipline and resources. If journalists’ reporting process (or other researchers data collection on journalists’ behalf) includes data collected and stored on IP-connected devices, or sent over IP networks, they multiply the vulnerabilities to interception.
In the wake of “the Snowden revelations” the Committee to Protect Journalists published a short blog post which noted that intelligence agencies “technology, the will, and some very loosely written laws that allow them to snoop with impunity”. (They weren’t the only ones. Indeed, the Tow Center Director, Emily Bell, has been quoted on this theme since at least 2011.) That post was written in the context of live human sources, but the principle seems applicable for electronic sources as well. If a sensor is collecting personal or potentially harmful data, then that data may need to be redacted, deleted or anonymized, and, as implied by the vulnerabilities of transmission, that process may be best implemented at source.
Group 2: Building and Acquisition
Journalists who are deploying sensors need to get them from somewhere; whether they’re making them “in-house”, commissioning custom devices, or buying off the shelf. In the case of built or commissioned sensors, media organizations may encounter the following legal and ethical considerations — some of which may be familiar to existing product development teams in media organizations.
Open and Closed Source, IP and Patentable Technology
Builds incorporating open source technology (whether software or hardware designs) may incur obligations to make subsequent products open source, or credit the original authors. In all cases, whether the license is GNU, GPL or any other variety of open source, the obligations are spelt out in the license and normally packaged with the source.
Journalists may also want to consider whether making their own work open-source (and therefore reusable by other parties) is consistent with their mission and goals. The Knight News Challenge, which has funded many innovative journalism projects including DocumentCloud and Ushahidi, requires that grantees open-source their code. Doing work under open source licenses also aligns with the principle of ‘Show Your Work’ which many news apps developers believe improves their journalism by making their decisions and processes transparent and replicable.
In a similar vein, when contracting work to supplier, if the work includes hardware design or original software development, a well-written contract should specify the ownership of the designs, the physical products, and any resultant patentable technology.
New Hardware and Communications Regulators
In the USA, the FCC needs to approve new electronic devices when they are brought to market. They are trying to mitigate the risk of interference between all the devices that involve electro-magnetism; from WiFi devices, to radios, to garage door openers, walkie-talkies and cellphones.
Sparkfun, the DIY electronics resource, has an extensive primer on the FCC regulations as they apply to small makers and prototype builders. According to Sparkfun, people making fewer than five devices for personal use run very little risk, but beyond that, journalists may want to consult their own legal resources. Legislation will obviously be different in other jurisdictions.
Group 3: Deployment and Operation
When journalists deploy sensor-carrying devices into the field, especially remotely or autonomously, they will attract different risks and liabilities.
Property Rights & Trespass
Sensors which monitor or record environmental phenomenon (whether light, sound, radiation, air “quality”, water “quality”, temperature, humidity, vibration) need to exist somewhere, sometimes temporarily, sometimes over a long period. While many citizen science projects have recruited participants to place sensors in their private property, journalists whose reporting plan requires samples to be taken from specific locations (so as to collect representative data) may have to contend with laws restricting their access to a given spot. (Leaving aside the risks involved in leaving electronic equipment to the mercy of the weather and curious passers-by.)
While many semi-public places like airports and subways forbid cameras and recording devices, many news photographers will be aware of police and private security guards over-reaching their authority to prevent recordings in a variety of spaces.
Personal Injury & Property Damage
Once a sensor is in a space (whether directly controlled by a journalist or running autonomously), it may pose risks to other people or their property. Many cities demand that newspaper companies insure their street vending boxes against personal injury and property damage. It seems reasonable to assume that sensor installations could either attract similar requirements and even if they don’t journalists would be wise to reduce their financial exposure to “trip and fall lawsuits”.
When WNYC’s data team was risk-planning a project to sense air temperature throughout the New York City subway system, one of the significant damage risks they analyzed was less prosaic; they didn’t want to be responsible for a passenger seeing an unusual electronic device and calling in a bomb scare — thereby shutting down a whole train line for hours while the scare was investigated.
Restricted Sensor Types
In June 2012, the Chinese government asserted that the US Embassy was “contravening relevant environmental protection rules” by monitoring and publicizing air quality in Beijing, according to The Guardian. Earlier, in the lead up to the 2008 Olympics, an AP/ MIT Civic Media Lab project smuggled air quality sensors into China disguised as camera equipment. In both cases the legality of importing and setting up those sensors is unclear, however in discussions that the Tow Center has had with other journalism organizations planning drone and sensing projects, reporters with experience entering more restrictive countries have suggested that unidentified electronic equipment (or even curiosity-piquing equipment) is at risk of being arbitrarily confiscated — whether there’s a legal basis for doing so or merely a perceived threat to the government.
Data Transmission & Storage
Remote sensors that wirelessly transmit data use some kind of radio bandwidth; be it cellular, WiFi, or some other frequency. As noted in the section about hardware regulation above, the FCC licenses specific spectrum for specific uses.
Cable and wired ISPs, as well as cell companies have sometimes restricted professional uses of consumer-grade subscriptions; they are mostly aimed at running servers on home internet plans but they are often phrased as ‘fair use’ and give more control to the ISP than the buyer.
These considerations should be cross-referenced with the privacy and surveillance issues; data storage and transmission are areas of vulnerability for hacking and eavesdropping.
Group 4: Accuracy, Interpretation and Communication
Journalists’ ethical codes place a premium on accuracy. At the top end of the news market, companies’ business models are built on credibility; audiences will visit, follow, pay attention (and even pay money), because they trust that the information they are getting is accurate, fair, meaningful, digestible, relevant, significant and doesn’t leave out important data.
The personal consequences of getting a story wrong might swing from damage to a journalist’s personal reputation (with possible career consequences), to weakening their organization’s brand and onwards to legal action.
The traditional ideal of journalists’ role — holding powerful forces to account and giving people the information they need to participate in civic life — is also predicated on distributing accurate information.
Newsrooms that have maintained their traditional values of accuracy as they practice more data journalism have developed deep expertise in new (for them) processes, including statistics and sharing their methodologies. If journalism does start to work more closely with sensors, it seems likely that newsrooms will need to acquire even more new skills, and a subtle understanding of the epistemological demands of different types of projects. My colleague Alex Howard has written about ethics in the context of data journalism in a separate article.
Collection, interpretation and communication
The subtleties of using sensors to produce news data would seem to have implications for how journalists plan data collection, interpret the data, communicate what they know (and what they don’t know).
The EPA’s guidance for good environmental data collection calls for precision, accuracy, representativeness, completeness, and comparability. They also note that equipment should have an adequate detection range for whatever is being measured. Those characteristics of a good data production should be relevant beyond environmental sensing; they can apply to sensing humans and objects.
Journalists can represent the level of confidence in the quality of their data throughout their process — right up to how they present it: The Washington Post’s investigative story ‘ShotSpotter detection system documents 39,000 shooting incidents in the District’ used data from an array of microphones commissioned by the city police force. The journalists incorporated the nuances of the data collection process into their story: they explained the limitations of the system’s accuracy, its biases, and designed their visualizations to portray the fact that their samples did not cover the whole city; just areas which were previously understood to have lots of shootings.
Relative, Absolutes and Calibration
Newsworthiness is often found when reporters can point to broken laws or demonstrate a problem; health risks for example. In the context of sensors this would make calibrated, absolute data particularly valuable.
While it is relatively easy to see relative changes within data collected by a single sensor (fine particulate matter or temperature, for example), unless a journalist has absolute values in that data, they will struggle to compare it to known health effects or legal standards.
(Journalists who are working on stories about toxicity and human health may also need to review exposure science, which describes how health outcomes are affected by the intensity of stressors and the amount of contact a person has with the potentially harmful substance.)
Different traditions of truth
The skills that can be applied to sensors in journalism come from a diverse range of fields; from humanities through the sciences. Generally, integrating different cultures into a shared endeavor works best when the differences are acknowledged, understood and negotiated. This will probably apply if scientists and journalists work together on sensor projects.
The phrases ‘news is a first rough draft of history’ and more latterly, ‘data journalism is social science on a deadline’, acknowledge that journalists often work within severe time constraints. They are sometimes required to cover a broad range of topics (especially in comparison to scientists). The traditional mainstream media is often communicating for an inexpert, easily distractible audience — if readers don’t fully and correctly understand the phenomenon being reported, it may not have a direct impact on their lives. These forces have had implications for how ‘truth’ is gathered and how it is presented.
By contrast, scientists publish within a system where deep specialization is the norm, their work is systematically peer-reviewed for accuracy and significance and there are direct adverse career effects for getting the information wrong. Meanwhile more of the ‘audience’ has a higher stake in consuming accurate information and getting a full, nuanced comprehension. Scientific publication takes longer, corrections and retractions are rare (although not unheard of), however the information does not spread as far.
When these two types of professionals collaborate on a project, they may need to examine and discuss their ingrained perspectives about clarity, specificity, generalizations and nicety, indeed the entire knowledge production and communication process.
Goals of Illustration versus Evidence versus Engagement
Even as limited as the field has been thus far, sensor journalism projects have already had a diverse range of goals, with corresponding differences in how accurate (or comparable, or reproducible) their data needed to be.
The WNYC Cicada Tracker achieved good participation; bridging the gaps between citizen science and journalism. It deepened the bond between media organization and its existing community, accessed a new community (makers) and helped educate participants in an aspect of biology and home electronics.
The New York Times’ ‘How To Win’ motion-capture work in anticipation of the 2012 Olympics used sensors to produce incredibly rich 3D illustrations. By translating movement into data it allowed their graphic artists to show audiences unfamiliar and insightful views of elite athlete’s movements.
Two USA Today investigative stories aimed to produce data that could be used to suggest health risks. The soil contamination data that USA Today produced for their ‘Ghost Factories’ project, and the air quality data they used for the ‘Smokestack Effect’ project gets closer to evidence-grade. They partnered with universities’ health departments (Johns Hopkins, University of Maryland, University of Massachusetts) to ensure their methodology was as sound as possible.
When I’ve discussed this idea — that different projects’ goals could mean different standards of accuracy — some journalists have been wary: It’s not immediately clear that different standards of accuracy are acceptable in journalism.
Next…
The ideas above call out for your responses. Are the groupings right? Are we missing issues? Are we applying traditional values in appropriate ways? Are we characterizing ideas about accuracy and privacy in a coherent way? Let us know in the comments, or write to ferguspitt@columbia.edu
Fergus Pitt is a Tow Fellow working on the Tow Center’s Sensor Newsroom Project at the Tow Center for Digital Journalism. The Sensor Newsroom Project is a project made possible by generous funding from both The Tow Foundation and the John S. and James L. Knight Foundation. The goal of the Sensor Newsroom Project is to explore how recent advances in sensor networks, citizen science, unmanned vehicles and community-based data collection can be used by a new generation of sensor journalist to move from data analysis to data collection. Follow Fergus Pitt on Twitter @Fergle. To learn more about the Tow Center Fellowship Program, please contact the Tow Center’s Research Director Taylor Owen: taylor.owen@columbia.edu.
