Citizen Science and the Politics of Environmental Data

In this commentary for the special issue on sensing practices, I reflect on some aspects of the politics of data collected as part of citizen science efforts to monitor environmental contamination. This is a sketch for a more extensive analysis, as well as a provocation raising questions about a particular aspect of the politics of data from sensing practices of citizen science—the extent to which these data help make environmental hazards more publicly visible. I pose these questions based on comparing one approach to citizen science developed by the volunteer network Safecast in Japan following the 2011 Fukushima Daiichi nuclear accident, with what could retrospectively be labeled a kind of citizen science—an independent approach to monitoring radiological contamination undertaken by the Belarusian nonprofit Institute of Radiation Safety, “Belrad,” in the different historical and cultural context of Belarus following the 1986 Chernobyl nuclear accident.

Safecast, organized to monitor radiological contamination after Fukushima, is arguably one of the best-known citizen science projects in the world (Abe 2015). Belrad has provided much of the publicly accessible data documenting the scope of contamination in Belarus, a former Soviet Union republic most heavily affected by the Chernobyl fallout (Kuchinskaya 2014). Belrad and Safecast were established (and continue to work) in different historical, political, and cultural contexts. Among other things, the Fukushima accident happened after the emergence of the Internet, social media, and cloud-based tools, with their potential for crowdsourcing, amplifying voices of the affected populations, and supporting new kinds of citizen science. Because the contexts are so different and, correspondingly, Safecast and Belrad demonstrate different approaches to data collection and assumptions about data, comparing these approaches might help us ask questions about the politics of data produced by citizen science.

My approach to the politics of data is based on my earlier work (Kuchinskaya 2014), where I argued that imperceptible hazards, such as radiation, need to be made observable and publicly visible (i.e., publicly recognized as a hazard). How radiation is represented matters, and the production of in/visibility is relative: some discourses, practices, and conditions render hazards more visible, while others, in comparison, render hazards less visible and potentially even nonexistent as a social issue. Public invisibility of environmental hazards does not necessarily mean that there is no danger but that those who are most affected are ignored and disempowered in the process of knowledge production.

Much of the analysis in my book focused on how radiation risk in Belarus was ultimately rendered less visible to the public, yet I emphasized the unique trajectory of that disappearance. Countless other imperceptible hazards, including those in Western contexts, are continually being made invisible by the industries that produce them and that are, in turn, aided by administrative bodies that do not regulate them. What is distinctive about the disappearance of the consequences of Chernobyl in Belarus is the historical waves of public invisibility and visibility of Chernobyl’s consequences. What stands out especially is the eruption of the public visibility of Chernobyl in the last years of the Soviet Union (1989-1991): extensive media coverage of Chernobyl, the passage of Chernobyl laws, and the establishment of several dedicated research institutes. The public visibility of Chernobyl increased dramatically before the collapse of the Soviet Union (and the turn to market relations), and it was made possible by the intensified glasnost and the cracks in the political regime of the Soviet Union (Kuchinskaya 2014). As I describe later, Belrad’s independent approach to data collection and analysis emerged during that historical period, inheriting particular assumptions about the politics of data.

“Citizen science” is a term that has different meanings for natural scientists and for science and technology studies (STS) scholars (Lave 2012). The former typically define it as volunteers “collect[ing] and/or process[ing] data as part of a scientific enquiry” (Silvertown 2009, 467). Volunteers’ participation is enabled by new cloud-based tools and the view of the public as a source of cheap creative labor. For STS scholars, the term implies activist, bottom-up practices aimed at democratization of science, often in the context of the production of local environmental knowledge (Lave 2012; see also Irwin 2015). I share the STS concern with the democratization of science and grassroots activism, although I propose to examine what particular practices of citizen science mean for bringing public attention to imperceptible hazards and their effects. In the sections that follow, I briefly consider radiation monitoring by Safecast and then contrast these efforts with the post–Chernobyl radiation monitoring by Belrad. I conclude with questions about citizen science, environmental data collection, and the production of public in/visibility of hazards.

Safecast, an international network of volunteers, was established soon after the Fukushima nuclear accident in response to inadequate government data on the scope of contamination. The goal was for volunteers to collect their own data and publish it openly, with no access restrictions. To collect data, Safecast volunteers use low-cost sensors linked to GPS units, put them on their cars, and drive around measuring radiation levels.

Safecast’s approach to radiological monitoring is different from traditional approaches that rely on using state-certified, calibrated, and expensive equipment operated by trained experts. Safecast’s model was shaped by some notable international technology figures, including a number of US technology entrepreneurs, as well as Joi Ito, the Director of the MIT Media Lab and one of the cofounders of Safecast (Abe 2015). Volunteers at the Tokyo Hacker Space developed cheaper mobile monitoring devices using open-source standardized hardware and open-source software, and the design was improved through iterations and field experiences. bGeigie, the device ultimately developed by Safecast, “enhanced the ease-of-use of the system, basically allowing ‘set and forget’ operation by even non-technical operators” (Brown et al. 2016, S85). Interestingly, volunteers often assemble their own bGeigie from a kit; the cost of the kit and the technical skills required to assemble it serve as a kind of self-selection for “relatively technically adept and motivated volunteers” who are more dedicated to data collection in the longer run (Brown et al. 2016, S90). Collected data are “licensed with a Creative Commons 0 license, allowing anyone anywhere in the world to use the data for any purpose” (Hemmi and Graham 2014, 835). The methodology for data collection and production of maps is posted online, which enables feedback from experts (Brown et al. 2016).

In short, Safecast’s approach emphasizes crowdsourcing, open-source, and open-data methodologies. Not only does this approach exemplify the best practices of citizen science, it also seems to be well suited to making the scope of contamination more publicly visible through greater coverage and more transparency in data collection. Yet one aspect of Safecast’s approach complicates how we might view Safecast’s role in making post–Fukushima radiation more publicly visible. Safecast explicitly avoids advocacy and describes itself not as “anti nuclear, or pro nuclear” but as “pro data. Data is apolitical” (Safecast blog quoted in Abe 2014, 5). Safecast provides only “raw data,” that is, data without an explanation of what it means for radiation safety. Radiation protection is an area of “deep-seated controversy,” and Safecast appears to try to avoid accusations of bias or having a hidden agenda (Brown et al. 2016, S93). This neutrality is attractive for at least some volunteers (Abe 2015); it also protects Safecast’s “open discussion process” from becoming “overwhelmed by disagreement about the data interpretation,” and it potentially allows the data to reach diverse communities of data users (Hemmi and Graham 2014, 839).

Safecast is not unique in their separation of data collection and advocacy. Similar practices are described by Kimura (2016) for post–Fukushima citizen radiation-monitoring organizations (CRMOs) focused on measuring radiation in food. Like Safecast, many CRMOs “seemed to prioritize producing data over challenging the utility and nuclear industries and pro-nuclear government” (Kimura 2017, 2-3). Kimura noted that the work done by these organizations is inherently political; they are working to increase awareness in the overall climate of policing radiation concerns. Yet most CRMOs shy away from interpreting whether measurements are dangerous for their clients. Kimura gave several reasons for this unwillingness, including the need for citizen scientists to position themselves as credible and the impact of neoliberalism, in which managing risks becomes affected individuals’ own responsibility. Also, many of these citizen scientists are women, and they face additional challenges to their credibility when raising risk-related concerns, which already are harshly policed as irrational and detrimental to economic recovery of the affected communities.

To appreciate what avoiding advocacy and presenting data as neutral might mean for making radiological contamination publicly in/visible, consider a different approach. Belrad’s work is based on interpreting radiological concepts and data as inherently political. Belrad was established in 1990, during the period of Chernobyl’s exploding public visibility. Oppositional intelligentsia in Belarus, including scientists, demanded declassification of Chernobyl documents, called attention to unrecognized contamination, and protested the “radiosecrecy” approaches of Moscow and its top nuclear experts (who argued that post–Chernobyl radiation exposure would have no observable effects for the general population and further radiation protection measures were unnecessary). In that context, references to the “expertise” of the leading Moscow scientists and the “objectivity” of their approaches had clear political stakes. When local Belarusian scientists proposed a more cautious approach to radiation protection, this approach was inseparable from their political views and their denunciation of Moscow’s attempts to obscure the scope of Chernobyl’s consequences.

Belrad’s view that radiation protection was inherently a matter of political decision-making and advocacy came from its founder, Vasily Nesterenko. Nesterenko had previously served as the director of the Institute of Nuclear Energy of the Academy of Sciences of Byelorussian Soviet Socialist Republic (BSSR) and the chief engineer of the mobile nuclear plant “Pamir.” He had gotten into trouble with the Communist Party leadership for writing a letter to the local government in which he called attention to the scope of radiological contamination. The establishment of Belrad was also supported by other key oppositional figures of the time, including nuclear physicist and dissident Andrey Sakharov and writer Ales Adamovich.

Nesterenko defined Belrad’s objective as “total control over individual doses” (Kuchinskaya 2014). He explicitly acknowledged that radiological contamination was too expensive a problem for the Belarusian state. For him, this meant that radiation protection should prioritize the most vulnerable populations such as children in the most affected rural areas. Belrad relied primarily on a relatively small number of experts and staff, and, compared to Safecast, their efforts were more difficult to scale. Yet the scope of their activities has been remarkable. It included design and production of devices, monitoring of individual doses with whole body counters (WBCs), and testing of food samples. At some point, Belrad was running a network of local centers for collecting samples of food in the most heavily affected areas; this work included training local community members. Belrad later developed capabilities for testing strontium-90 in food and as internal accumulation in people (strontium-90 is a β-emitter that requires a more complex, labor-intensive, and expensive process of testing). In short, Belrad staff worked to identify what was needed to document (make visible) the scope and character of radiological contamination—filling the gap in radiological monitoring of the Belarusian state, developing and adjusting own capabilities, and focusing on the circumstances of the most affected and vulnerable populations.

Belrad’s data collection was inseparable from its advocacy. Its staff worked to educate local residents about reducing their exposure, educate local administrators and thus facilitate concrete decontamination measures, and even affect radiation protection policies of the Belarusian government. For example, Belrad acquired mobile WBCs, transported them in minivans, and tested schoolchildren in affected rural communities (who could not be easily reached by other radiation-monitoring initiatives and yet were known for consumption of potentially contaminated products from private lots and local forests). The data were used to challenge the official status of some areas that were no longer recognized as contaminated (Kuchinskaya 2014). In this and other cases, Belrad lobbied local and national authorities based on juxtaposing its own data with state data or existing radiation protection norms and thresholds. Such lobbying required the use of state-certified equipment and techniques (as well as expert credentials and opportunities to enter into dialogue with regulatory bodies).

In summary, Belrad’s practices were grounded in the radiological and political assumptions of its experts: they shared the cautious radiation protection approach proposed by the Belarusian scientists in the last years of the Soviet Union (and adopted into the 1991 Chernobyl laws in Belarus). From their perspective, chronic exposure to low-dose radiation was not harmless. Belrad’s practices fused data collection and advocacy, and they relied on the use of state-certified equipment and techniques in order to engage with state and local authorities. Especially striking in Belrad’s work—the work performed by a small group of independent experts—is the ambitious comprehensiveness of their approach to radiation protection, which in practice meant prioritizing the most vulnerable populations.

Culturally and historically different circumstances in Japan allowed for the emergence of many more citizen organizations that could do the technical work of radiation monitoring, thus scaling up the work of experts. Safecast alone has generated vast amounts of data. Safecast’s open-data and open-source approach is technologically and methodologically exciting. It promises to challenge the established hierarchies of expertise with its cheaper devices and the use of crowdsourcing. At the same time, the design of the project is not a guarantee of its effects (Kimura and Kinchy 2016). And comparing the practices of Safecast and Belrad raises questions about the relationship between data generated by citizen science and making environmental hazards publicly visible.

The connection between generating environmental data and public attention to radiation as a hazard is not necessarily straightforward; it is a question of the politics of data (Gitelman 2013; Kuchinskaya 2013, 2018). The politics of data could be explored through questions similar to Belrad’s interrogation of both their own and state-produced radiological data, focusing on how data are used, what they render invisible, and whether and how they serve the most vulnerable and affected populations.

First, there is a need to document the actual uses of data produced by Safecast and CRMOs, especially when and how the data produced by citizen scientists are used to help bring about positive change (Conrad and Hilchey 2011). It is not enough for citizens to produce their own data; it matters how data are interpreted and who controls how that interpretation is undertaken (Ottinger 2010).

There are also questions around the general usability of data and the conditions that would allow citizen science data to be used for advocacy. The work of Safecast and CRMOs is often delegitimized by establishment experts for not using state-certified, properly calibrated (and expensive) equipment (Abe 2015; Kimura 2016). Safecast challenges this traditional approach and argues that numerous measurements using identical low-cost devices allow for “increased statistical accuracy over time through larger sample size and redundancy” (Brown et al. 2016, S94). From Safecast’s perspective, their success “provided an important proof-of-concept” (Brown et al. 2016, S94) and other similar initiatives followed. Still, this uneven and limited acceptance of citizen science data needs to be explored, potentially in the context of the overall growing recognition of citizen science by the current neoliberal knowledge regimes (Lave 2012). If the data are deemed usable, the question is when, by whom, and with what effect in terms of the production of public in/visibility of radiological contamination.

Second, the usefulness of citizen science data generally depends on transparency about the underlying assumptions (Silvertown 2009). Safecast’s position assumes that the data will find its audience; that there are few, if any, assumptions built into the data; and that, given the overall transparency, the data will be usable for current or future knowledge production goals. Yet, as STS scholars have insisted, “‘raw data’ is an oxymoron” (Gitelman 2013). The production of data is necessarily shaped by particular assumptions and practices, and the question is: What is left out or made invisible in Safecast’s data? (e.g., it might be that some years after the accident, testing contamination in food is more important than monitoring radiation in the air.)

The approach adopted by Safecast does not guarantee the production of data useful for the most affected and vulnerable communities and sensitive to their specific circumstances. Yet, as we have seen in the example of the work of Belrad, this sensitivity might be crucial for making radiological contamination and its effects more publicly visible. The politics of data needs to be examined even when the data are vast and the methods of data collection are democratic and powerful. The question we are left with after considering post-accident knowledge production practices is not about the objectivity of data but about how socially just particular methods and data are and whose interests they serve or do not serve (Kimura 2016; Kuchinskaya 2014).