Sensors and Journalism

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.1Journalists 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.