A (Working) Typology of Sensor Journalism Projects
If you’re ever feeling like life is too structured, you can always blame Aristotle.
The Greek philosopher laid the groundwork for a system of biological classification by observing animal traits and methods of reproduction. (Granted, his system was later rendered obsolete, but he still gets street cred for getting us started.) Then, I suppose, you could blame Carl Linnaeus, who improved on Aristotle’s work in Systema Naturae (1735) and introduced the modern method of formulating scientific nomenclature for all forms of life with regard to kingdom, phylum, class, order, family, genus and species (hello, 9th grade biology!). Thus, we have a systematic way of categorizing and understanding life itself.
But why classify in the first place? Why this human habit of indexing, compartmentalizing, and organizing in a world that inevitably gives into entropy and chaos?
Without beginning a philosophical harangue, my immediate answer is that we combat chaos and create categories for three main reasons:
- To make sense of the world to ourselves.
- To understand relationships between things and groups of things.
- To construct a way to communicate within and outside of a discourse.
The same logic follows with how I’ve been thinking about the diversity of projects spawning in the sensor journalism space as I (for lack of a better expression) try to make sense of it.
Sensor journalism — using sensors to generate data around journalistic inquiry — is a nascent practice and is, therefore, unavoidably in flux.
Through my research and advocacy for this work, one of my challenges has been figuring out how to contextualize the sensor-based projects I’ve heard about or been a part of. While practitioners are experimenting with new ideas around how sensor data can enrich, ratify, or break news stories, there comes a pressing need for a way to frame existing approaches.
I propose a working typology of sensor journalism projects in order to examine the work that has been done; tease out emergent patterns; and gain perspective on the field to anticipate its future trajectory.
[If you haven’t already, I highly suggest reading the Tow Center report called Sensors & Journalism, which does a brilliant job of highlighting important historical case studies of sensor-enabled journalism. An elision that I noticed, though, was mention of contemporary examples, a gap I hope to help fill in here.]
To begin, I’ll borrow from Charles Berrett’s definition of “sensor” in the Tow Center report to clarify what I mean by the term: a sensor is something that reacts predictably to the state of the world. This would include thermometers, cameras, chemical tests, accelerometers, among many others.
I’m also going to define “journalism” quite liberally to encompass any type of content that is generated — and in any format (i.e. photography, videos, audio, tweets, blogs, print, data visualization, etc.) — based on observations of the environment, and that includes reporting on one’s own self.
With today’s technology and within the realm of journalism, a sensor can be a type of “closed” or “open” technology. Closed technologies are constrained by an explicitly stated intended use and design (e.g. an arsenic sensor you buy at Home Depot), whereas open technologies are intended for modification and not restricted to a particular use or environment (e.g. an open hardware sensor you can build at home, based on a schematic you find online). Both types of technologies can be employed by people to generate data. Data from both types of tools (closed + open source) can also be combined for the purposes of establishing provenance. For example, an aerial drone can capture images of an area suffering from deforestation, which can then be compared with images taken from the ground with a camera phone for higher resolution of the damage.
Which brings me to my last introductory point. It’s important to consider who the agents are in the following types of projects, i.e. who is actually using the sensors. In some scenarios, the agents are journalists; in others, the agents are citizens; and in others yet, the agents are government organizations or private companies. This affects how the data are collected, who gets access to the data, and ultimately how the data are interpreted.
Below is a provisional typology of sensor journalism projects. It is not, by any means, perfect nor complete. However, it is how I’ve begun to classify projects in my mind, and I welcome you to add to it, tear it apart, or (if I’m lucky) adapt it for your own purposes.
Types of sensor journalism projects
Journalists using sensors
In this scenario, a journalist or small team of journalists would be trained to use a sensor tool, collect data in the field, analyze the data (either independently or with a collaborator), and report on or visualize the data.
In spite of controversial policy debates with the FAA about drone technology, some journalists are able to use drones to capture aerial footage of spaces to report on such things as crowds, traffic, and environmental health. Many drones are commercially manufactured, which would make them a form of closed technology.
An open source alternative would be something like Public Lab’s balloon mapping kit, which has been used for similar purposes, namely taking aerial photos for community mapping of spaces and identifying environmental hazards.
Both of these sensors can be used to achieve similar results, and newsrooms/journalists should be able to select which types of sensors would best match their needs for the story and issue.
People using sensors
Newsrooms can also engage their constituencies with sensor tools by either 1) putting sensors in people’s hands or 2) asking people to build sensors.
Projects like SafeCast put sensors in the hands of people. One can purchase the sensor online and build the sensor at home to independently measure radiation levels, as opposed to relying on a third party. This sort of distributed sensing became especially pertinent after the Fukushima-Daiichi earthquake in Japan. Instead of relying on government agencies to report back on their readings, people on the ground could deploy these sensors to get higher resolution on local radiation levels.
People can also build sensors themselves to gather data, whether prompted by personal inquiry or a news outlet. The classic example of this (which I may have exhausted at this point) would be WNYC’s Cicada Tracker, wherein listeners of the public radio station WNYC were invited to build a soil temperature sensor, take readings, and report the readings back to the station. The data news team visualized the self-reported data points and blogged about the process along the way.
Although there haven’t been many projects like this since, I’d like to point out the potential of more projects like it crystallizing due to the growing availability of low-cost technology and the growing DIY/open hardware community.
For example, the Open Water Project aims to “develop and curate a set of low-cost, open source tools enabling communities everywhere to collect, interpret, and share their water quality data.” Their RIFFLE (Remote, Independent Field Friendly Logger Electronics) sensor measures temperature, depth, and conductivity of water. The conductivity circuit in this sensor (nicknamed Coquí) is available as a kit and can be built at home. The data output can also be translated into audio, offering a different way to interpret sensor data.
The issue of data quality and sensor calibration comes up most frequently in discussions about putting sensors in the hands of people. It’s a fair concern, as distributed sensing often means less control over how the sensor is deployed, the context in which the data are collected, and how the data are reported. Contingent on for whom the data would ultimately serve, these factors may or may not matter. If the goal is to engage people in environmental sensing or to simply observe general trends (e.g. range of temperature readings in a local watershed), accuracy might not be a priority. However, if the intent is to map out chemical contaminants in a local area, which might freight social, economical, cultural and political impacts, this would be a completely different story. This then raises the issue of how we should evaluate distributed sensing initiatives: what matters most when it comes to sensors and their data, and in what order of priority? Data transparency? Data accuracy? Sensor calibriation? Adherence to established standards?
Here, I’d also like to point out once again that my definition of journalism, for the purpose of this context, expands beyond newsrooms and includes such forms as blog posts, tweets, and other other form of documentation of a process or issue. That said, blog posts or research notes that document the process of using these sensors would qualify as a form of journalism.
Existing sensor networks
There are also sensor networks already embedded in spaces around us. Most are controlled by city-level groups or private companies, but in some cases, journalists may be able to get access to these data streams for their stories. For example, the Tow report highlights the Sun Sentinel’s investigation on speeding cop cars based on sensor data from toll gates:
We've all seen it, and now there's proof: Police officers sworn to uphold our traffic laws are among the worst speeders…articles.sun-sentinel.com
The state of Utah recently announced the implementation of “smarter,” sensor-based traffic signals that will adjust in response to flow and help control congestion. Imagine if this data could be made public or available — how could it help raise awareness about common traffic patterns for city planners, commuters, etc., and how could journalists and storytellers help mediate the messaging?
Drive your Model T through a major intersection a hundred years ago, and you'd likely encounter a policeman directing…www.citylab.com
SensoredCity is a new Robert Wood Johnson Foundation-funded initiative to “design, develop and install a network of sensor-based hardware that will collect environmental information at high temporal and spatial scales and store it in a software platform designed explicitly for storing and retrieving such data.”
SensingCity in Christchurch, New Zealand, that endeavors to place sensors almost everywhere within the city to track, well, everything. It operates on three core principles, however: (1) all data are open and accessible to the public (2) never track individuals, and (3) track everything.
It remains to be seen to what ends sensor network data can be used for journalism or whether the data are useful, accurate, or compelling. Then the question of how to best make the data available and legible for a general audience also needs to be addressed. Nonetheless, the growing presence of sensor networks is identifiably another possible resource for information and data about public spaces.
A smorgasbord of sensors is already embedded in our smartphones. Potentially, this would enable anyone with access to one to report on data, given the availability and access to the proper software to do so. To illustrate, check out the Android developer blog breakdown of all the different sensors that are embedded in smartphones, which include (among so many other sensors) GPS, accelerometer, temperature, light, pressure, et al.
SeeClickFix is a civic app that allows users to report non-emergency issues. In Detroit, residents use SeeClickFix to report damaged water pipes that need repair to the City of Detroit Water and Sewage Department. The problems are usually mapped and logged in a news feed. Ideally, an app like this enables a platform on which citizens and cities can collaboratively identify issues.
StreetBump is an app that logs accelerometer data as bikers and drivers commute to help report on potential potholes and inconsistencies on public roads. Still, until data from these sensor-based apps actually influence a change, apps like this are merely a proof of concept. Case in point: at the time of writing this, there have been almost 40,000 bumps reported but zero potholes have been filled as a result. The questions of data quality, sensor calibration, and who is accountable for implementing changes should be explored more deeply.
Because software development tends to grow at a faster rate than hardware development, perhaps mobile sensor journalism use cases would be easier to prototype and evaluate. Another important point to keep in mind is that mobile sensing would be limited to communities of people who have access to smartphones, which may geographically or economically restrict which strategies could be deployed.
Remote sensing is carried out by sensors aboard various platforms: planes, boats and Argo floats. As is the case with existing sensor networks (above), these remote sensors are often controlled by organizations or corporations. However, it is possible to get access to the data that they collect, as is the case with weather data from geosynchronous satellites.
Ground-based instruments are also used, e.g. sun spectral radiometers that measure solar radiation. However, satellite remote sensing is capable of providing more frequent and repetitive coverage over a large area than other observation means (Yang, et al, 2013).
Satellite remote sensing has provided major advances in understanding the climate system and its changes. For example, satellite images provide data about the spatial patterns of sea-level rise, the cooling effects of aerosols in the atmosphere, carbon sinks, etc.
ForestWatchers is a volunteer computing citizen science project, a platform that enables anyone (locals, volunteers, NGOs, governments, etc.) to monitor selected patches of forest across the globe, almost in real-time. A volunteer would use the Crowdcrafting platform to classify images of deforested areas to help researchers code the data.
Another project that plans to use satellite remote sensing data is iSeeChange, a mobile engagement initiative to galvanize people to self-report local weather/climate observations on social media. The next phase of the project syncs self-reported observations with sensor data from NASA’s OCO-2 satellite and BEACO2N (Oakland, CA) air quality sensor data. Findings will be reported on local media stations and their distribution channels.
Satellite remote sensing isn’t perfect, as there are still some technological challenges in assessing long-term trends of climate variables with remote sensors; however, it is another form of gathering data that could potentially support journalistic inquiry.
Wearables and data donation
Whether you are oblivious to or wary of the potential privacy issues concerning wearables like the FitBit, JawBone, the iWatch, and Google Glass, it is no mystery that these devices have grown in popularity. They are marketed as trendy and up-and-coming methods of computing, while the hazards of data leakage hardly ever make headlines.
At the time of writing this, it is possible to personally identify people wearing FitBits via their data broadcast streams.
On the one hand, this means that it’s possible for someone (or more than one person) to co-opt an individual’s data stream for secondary use. However, if an individual decided that they wanted to donate their data for some purpose, this could be another way of generating and getting access to sensor data. Beyond healthcare, another example might involve an environmental sensor designed for personal use, like the Air Quality Egg, which measures outdoor air quality (but can also be used for indoor air quality). The community around this sensor already maps where active sensors are and shares the data that are reported from these sensors.
Data repositories like DataHub and Xively already exist and provide and infrastructure for potential data donation and sharing. There are also people who have identified data donation as a solution for health care research. However, the question of how to obfuscate personally identifiable data, then to what degree to do so, is still up for legal, ethical, and philosophical debate. Although there has been an academic survey of health-tracking wearables, a more robust study on data privacy (i.e. how personally identifiable your information is) regarding other wearables should be conducted to better understand the landscape.
If it were possible to donate data streams or data sets safely, this might be another venue for research, journalism, storytelling, and pedagogy.
I don’t expect this typology to be widely accepted nor do I claim that it’s accurate. My hope is that anyone who reads this will help me recognize other ways of grouping projects that I haven’t quite thought of yet or help point out weaknesses or inconsistencies in these structures. Regardless, I appreciate you making it this far!
My default reaction with anything involving sensors, storytelling, and civic action is a positive one, so I am increasingly aware of my own biases and assumptions in my examinations of sensor journalism. However, I am beginning to consider a more critical view of the field in my own research in hopes of identifying where current gaps might be. There is much work and more sensing to be done yet.
Boylan, Michael. “The Biological Practice: Outlines of a Systematics.” Marymount University. 16 September 2014.
Drone Journalism Lab. http://www.dronejournalismlab.org/
“Mapping the World Via Cell Phones.” OpenSignal. Published Feb 12, 2013. Accessed September 18, 2014.
Pantelopoulos, Alexandros & Nikolaos G Bourbakis. “A survey on wearable sensor-based systems for health monitoring and prognosis.” Systems, Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions. Vol. 40, Issue 1. Published January 2010. Accessed 19 September 2014.
Pitt, Fergus (editor). Sensors & Journalism. Tow Center for Digital Journalism. Funded by the Tow Foundation and John S. and James L. Knight foundation. 2014.
“What is Open Journalism and What is Its Appeal?” Open Journalism Report of the WorldAssociation of Newspapers and News Publishers, 2012, www.wan-ifra.org/articles/2012/07/10/
Yang, Jun et al. “The role of satellite remote sensing in climate change studies.” Nature. Vol. 3, No. 10. Published 15 September 2013. Accessed 19 September 2014.