Structural Ignorance of Expertise in Nuclear Safety Controversies: Case Analysis of Post-Fukushima Japan

Abstract

This study sheds new light on nuclear risk governance from a sociological perspective by analyzing cases of post-Fukushima controversies on nuclear safety and nuclear emergency preparedness in Japan. By critically analyzing how the three risk-related concepts and methodologies, namely, probabilistic risk assessment, safety goals, and the System for Prediction of Environmental Emergency Dose Information, have been interpreted, implemented, and/or abandoned before and after the Fukushima accident, this study identifies three common features that characterize Japan’s nuclear risk governance: avoiding critical conflicts, proclivity toward automated decision making, and strategic overlooking of “uncomfortable knowledge.” These features all involve ignorance of the dynamic nature of safety where addressing uncertainties, heterogeneous knowledge, and incommensurable values can be key for continuously reviewing the existing edifice of safety. By elucidating why such ignorance persists in Japan despite the post-accidental drastic reform, the authors both articulate the deep-rooted structure that underlies it and reflects the societal and historical context, and eventually conceptualize this ignorance as “structural ignorance” of expertise in nuclear safety controversies and policy processes. The results also provide direction for further research to solve this structural problem.

Keywords:

• Structural ignorance
• Fukushima nuclear accident
• risk governance, uncertainty

I. INTRODUCTION

The last ten years have witnessed a massive number of accident investigation reports and scholarly analyses around the Fukushima-Daiichi nuclear accident and have drawn far too many lessons to mention here. Reflecting on them not fully but at least partially, the Japanese nuclear sector has expended considerable effort toward improving safety and preventing future accidents, including upgrading the safety measures of nuclear facilities, expanding governmental and industrial investment for safety and risk research, and political reform covering widely from legislative framework to regulatory governance. A symbol of post-accidental drastic change must be the dissolution of the previous major official institutions that used to determine the Japanese nuclear regulation by the central government: Nuclear and Industrial Safety Agency (NISA), the primary regulatory body; Nuclear Safety Commission (NSC), the regulatory advisory body; and Japan Nuclear Energy Safety Organization, the technical support organization (TSO) of the regulatory body. Their functions were consolidated by the new “independent” regulatory body, the Nuclear Regulation Authority (NRA), which was established on September 2012. As an “Extraordinary Organ” of the Agency for Natural Resources and Energy (ANRE) with strong interests in promoting nuclear energy use, the NISA’s lack of independence had been widely criticized both domestically¹ and abroad.² Consequently, the NRA was created as an “Affiliated Organization” of the Ministry of Environment in view of “the separation of nuclear regulation and promotion”³ representing a clear break from the past.

However, some social scientific studies have critically pointed out the “changelessness” of the Japanese nuclear program, for instance, in terms of policy process around nuclear energy use.^4–6 Instead, some scholars argue that the situation is worsening compared to the pre-accident era. For example, Taniguchi⁷ insists that the deficits of nuclear regulatory governance worsened after the substantial reform of the nuclear regulatory system. Juraku⁸ reveals that the social learning process through the official accident investigation functioned as a “forgetting” process instead of meeting its original objective.

Reflecting such alleged changelessness of the Japanese nuclear program, which here involves the nuclear industry, regulatory agency, governmental organizations, institutional frameworks, policy, and decision-making processes as well as relevant experts, this paper addresses the following questions: Can it be said that the ways to construct nuclear safety in this country have changed after the accident? Is there any critical issue around nuclear safety still remaining overlooked or ignored? If there is in fact any, why does such ignorance perpetuate despite the seemingly drastic reform of nuclear risk governance? By analyzing how the conceptual frameworks and scientific methodologies relevant to nuclear safety and emergency management have been accepted, institutionalized, or disregarded, this paper analyzes the common mechanism that prevents us from effectively learning from the accident and from the accumulated wisdom in the international scholarship and technical and policy practices. We assume that some deep-rooted problems obstruct the fundamental change to overcome ignorance, and analyze them in the following sections.

II. SCOPE AND METHODS OF THE STUDY

In the following sections, we introduce the case analysis of three post-Fukushima controversies and reforms associated with risk assessment and management: probabilistic risk assessment (PRA), safety goals, and the System for Prediction of Environmental Emergency Dose Information (SPEEDI). These three topics have caused active technical and social controversies among the experts and stakeholders in civil society both before and after the Fukushima accident. However, we have not seen substantial contributions of these safety concepts and methods toward the improvement of nuclear risk governance, as the studies previously mentioned show. Thus, it would be worthwhile to analyze the mechanism that caused those controversies around nuclear safety and risk.

In all three cases, the authors try to detect the shortcomings in the adoption, interpretation, and implementation of the key concepts by comparing the past and contemporary international discussions regarding them. It is by no means intended to accept international expertise uncritically and to call for the direct import of “established” toolkits to Japanese nuclear risk governance. What we would like to criticize here is the process that hollows out the substance of those concepts, which includes both advantages and disadvantages, significances and limits, of course, and reduces them to pseudo assurance on the risks of the nuclear program or to false consensus on controversial and delicate social issues. It is deeply problematic if the Japanese policy process fails to acknowledge contemporary expertise relevant to the key concepts, while the government, industry, experts, and relevant actors should have made many substantial contributions to preventing the next tragedy by reflecting on lessons learned from the disaster.

To reveal the reason behind the problem suggested previously, case analyses were conducted as attempts at interdisciplinary collaboration between risk governance studies and the sociology of science and technology based on qualitative studies, mainly by document analysis and interviews with individuals involved in the topics, conducted by the authors, as listed in the Appendix. Regarding these three cases, the authors in their previous studies found that the “already known” problems among the relevant experts have either not been considered, or have been ignored in the post-Fukushima controversies and policy process.^9,10

After the descriptive case studies, we then extract the three common features observed across the cases at the intersection of science, technology, and society: avoiding critical conflicts and dilemmas, proclivity toward automated decision making, and strategic overlooking of “uncomfortable knowledge.” Finally, based on these findings, we attempt to conceptualize the mechanism of that phenomenon as “structural ignorance” of expertise, with reference to other works and conceptual frameworks in social scientific scholarship.

III. CASE STUDIES OF POST-FUKUSHIMA CONTROVERSIES AND REFORMS REGARDING NUCLEAR RISKS

III.A. Probabilistic Risk Assessment

As multiple accident investigation studies were run in parallel in the aftermath of the Fukushima accident, the failure to use PRA in nuclear risk management was recognized as one of the contributing factors to the disaster. For example, the Report of the Japanese Government to the International Atomic Energy Agency Ministerial Conference on Nuclear Safety critically noted that probabilistic safety assessment (PSA), used interchangeably here with the term PRA, had not always been effectively utilized by the nuclear industry and regulators in Japan before the Fukushima accident.¹¹ The report of the Atomic Energy Society of Japan (AESJ) also recommended the proactive use of PRA to be a key factor in addressing the risk of extreme natural events and severe accident.¹²

The Japanese nuclear sector can look back to over 30 years of domestic history of the development and utilization of PRA, contributing to several safety practices in the nuclear industry. These practices included the safety designs for new advanced light water reactors (LWRs) and fast-breed reactors around the 1980s, the identification of effective countermeasures against core damage and containment failures, and the establishment of accident management measures in response to the post-Chernobyl regulatory requests.

However, until the Fukushima accident, the utilization of PRA in Japan lagged behind the global trend of adopting a risk-informed approach, both in industry and regulation. For instance, for the development of PRA methodologies and their practical use in industrial settings, the U.S. nuclear community had deepened the discussion on introducing risk insights. Consequently, in the 1990s, the U.S. Nuclear Regulatory Commission (USNRC) clearly steered in a more risk-informed regulation via, to name just a few, the PRA Policy Statement in 1995 (Ref. 13) and Regulatory Guide 1.174 in 1998 (Ref. 14). Even in France, where the introduction of PRA had been less proactive, the significance of PRA had been gradually paid attention to through some pilot PSA studies by the Institut de Protection et de Sûreté Nucléaire,¹⁵ leading to the declaration of extensive use of PRA in the Basic Safety Rule 2002 by the French Nuclear Safety Authority.¹⁶ In Japan, although the then-nuclear regulatory agency and advisory body had persuaded the operators to implement PRA and conducted a number of studies on risk-informed regulation since the 1990s, it did not result in the fundamental reform of the ways to construct nuclear safety (see Table I) (Ref. 17).

TABLE I Efforts to Introduce a Risk-Informed Approach in Japan*

Year	Month	Event/Title of the Report
1992	6	ANRE: Implementing the Periodic Safety Review (requesting PRA not as mandatory requirement but as voluntary assessment)
	7	Ministry of International Trade and Industry: Course of Action for Accident Management Measures (requesting individual plant PRA and accident management measures)
1994	10	Ministry of International Trade and Industry: Study Report on the Development of Accident Management in the LWRs
2002	10	NISA: Evaluation Report on the Results of Developed Accident Management in the LWRs
2003	11	NSC: Basic Strategy for Introducing Risk-Informed Regulation for Nuclear Safety
	12	NSC: Interim Report of Consideration on Safety Goals; NISA: Implementing Periodic Evaluation of the LWRs (legalizing periodic safety review but PRA remained as voluntary assessment)
2005	5	NISA: Basic Framework of Risk-Informed Regulation for Nuclear Safety; NISA: Implementation Plan of Risk-Informed Regulation for Nuclear Safety
2006	3	NSC: Report on Performance Objectives of Commercial Light-Water Nuclear Facilities
	4	NISA: Basic Guidelines for Risk-Informed Regulation on Nuclear Power Plants (trial version); NISA: Quality Guidelines for PSA on Nuclear Power Plants (trial version)
2007	9	NSC: Challenges and Future Prospects for Course of Actions of Relevant Agencies for Introducing Risk-Informed Regulation
2009	6	AESJ: Challenges and Recommendations for Codifying Standards for Full-Scale Introduction of Risk-Informed Approach

*By the authors based on Saito.¹⁷

We will not dwell on the details regarding how PRA had been positioned in safety decisions and regulatory frameworks before 2011 owing to the existence of several investigative reports on the subject.^1,18 However, very little social science literature has analyzed the background and in-depth reasons for this limited use of PRA.

Since the Fukushima disaster, there has been a considerable debate about the introduction of PRA and Risk-Informed Decision Making (RIDM), both of which are construed as the “right solutions” toward compensating for the delay in ensuring preparedness against severe accident and external events. Experts and practitioners involved in this discussion often cite the U.S. response to the Three Mile Island accident, mentioning the practical way of utilizing the PRA method and the important role of the Institute of Nuclear Power Operations in facilitating industrial proactive activities beyond the regulatory requirement.¹⁹ They successfully arrived at the conclusion that led to the establishment of the Nuclear Risk Research Center at the Central Research Institute of Electric Power Industry for developing and employing state-of-the-art methods of PRA, as well as the Japan Nuclear Safety Institute to ensure that the Japanese nuclear industry pursues “the highest standards of excellence.”²⁰ On the regulator’s side, the NRA, a newly established regulatory body, made the licensees conduct a mandatory PRA and compile their results in safety analysis reports to evaluate the safety improvement of commercial nuclear power plants, reprocessing plants, and fuel manufacturing facilities.

Despite these efforts, it is still difficult to say that a series of institutional reforms have fundamentally changed the ways of constructing nuclear safety by incorporating risk insights into safety decisions as integral elements. Rather, the prevailing Japanese framework seems to institutionally insulate PRA from other safety considerations, contrary to the ideally declared RIDM concept, which requires the integration of heterogeneous knowledge, including the insights from a deterministic approach and PRA (Ref. 21).

For example, the new regulatory standards, proudly claimed by the NRA to be “the world’s severest regulation,”²² require a high standard of resistance against external hazards, such as earthquakes and tsunamis, as well as a number of additional countermeasures to prevent and mitigate severe accidents. This tightening of regulation, however, was not fully reflected by the risk insights offered by the PRA, which adhered only to the conventional deterministic approach.²³ In the current configuration, the risk insights from the PRA were merely used as “toppings” on a solid deterministic regulatory layer, not as a key to encouraging reconsideration of the basic assumptions of nuclear safety. For instance, the licensees are required to conduct individual plant PRAs when they evaluate the effectiveness of measures against severe core damage in the conformity review according to the New Regulatory Requirements. However, the results of the PRA will be referred to only “if accident sequence groups that cause a significant frequency or impact and are not included in the accident sequence groups designated by the NRA are identified.”²⁴ The effectiveness of countermeasures against severe accidents itself is required to be evaluated only by the conventional deterministic analysis of incident progress. As such, the NRA’s ways of referring to PRA is not systematically integrated into the regulatory decisions, but only done in a dispersed way.

Further, a strange demarcation between “regulatory compliance” and “voluntary safety improvement” (to continuously reduce risks lower than the mandatory requirements) renders obscure the significance of embodying the alleged RIDM concept. In contrast to the growing need to synthesize various considerations in the RIDM framework, the regulators still depend heavily on deterministic analysis, and PRA activities are only undertaken by the industry. The current ministerial jurisdiction boundary between the NRA and the Ministry of Economy, Trade and Industry (METI) epitomizes the previously mentioned issue; while the former has jurisdiction over the “world’s severest” regulatory standards,²² the latter oversees voluntary industrial efforts, in particular PRAs. This dichotomous institutionalization is also pervasive in the onsite organizational structure of some nuclear power plants; the department in charge of voluntary safety improvement is divorced from the department responsible for regulatory compliance.

Only if PRA comes into use in redefining the basic assumptions of the existing safety framework can the authors regard PRA as being essentially beneficial for nuclear safety. By closely examining the history of developing RIDM in the United States from a governance viewpoint, Sugawara²⁵ framed RIDM as a dynamic process of cultivating an understanding of nuclear safety via a presentation of competing arguments from both the deterministic and probabilistic approaches by the stakeholders. This dynamism is epitomized by long debates in the 2000s between the USNRC and the nuclear industry over a loss-of-coolant accident (LOCA). The conventional deterministic concept of LOCA and methods of safety analysis around the emergency core cooling system had been challenged by the risk insights obtained from PRA, which included a fundamental question on safety; namely, is this requirement an unnecessary regulatory burden or is it an indispensable conservatism in terms of defense in depth? By referring to heterogeneous insights from both the deterministic and probabilistic approaches, RIDM can challenge the traditional safety framework and facilitate a discussion on what really matters in nuclear safety. It is only by squarely addressing the conflicts between both approaches that we can evolve our safety philosophy and practice. In this way, PRA can provide chances to shake and reconstruct the existing edifice of nuclear safety before a major accident happens.

From this point of view, the idea that introducing PRA methodologies can itself be a technical solution for preventing the next nuclear disaster is much too simplistic and opportunistic. Taking into consideration Downer’s criticism as the very nature of risk assessment systematically overstates such an assessment’s completeness and certainty,^26,27 such a naïve understanding of PRA will place reactors and our society at risk. The fruit of PRA cannot be enjoyed without a dynamic and reflexive process of revisiting what nuclear safety constitutes. The roles of PRA in this sense still remain ignored in Japan.

III.B. Safety Goals

An oft-quoted excuse for the delay of introducing PRA in Japan is the issue of “public acceptance”; that is, the Japanese public may not understand “risk thinking.”²⁸ It is not unique to Japan, however, that there exists a discrepancy between technical and quantitative risk assessment and public perception of risk. In particular, the nuclear risk issue at the interface of science, technology, and society has been one of the main targets of risk scholars and practitioners, and has resulted in the development of risk knowledge and practice.²⁹ The establishment of “safety goals” is a known example of such efforts. Many countries and international organizations have addressed the issue of “how safe is safe enough,” a conventional but fundamental question regarding nuclear safety, by establishing safety goals.

For example, the Safety Goals Policy Statement published by the USNRC in 1986 comprised qualitative and quantitative goals that attempted to provide an overall framework regarding the risks acceptable to individuals and society.³⁰ It reflected the technical advancement of PRA since WASH-1400 as well as the growing needs for formulating a safety goal triggered by the Three Mile Island nuclear accident.³¹ Meanwhile, the United Kingdom’s approach is less prescriptive, with the principle of as-low-as-reasonably-practicable (ALARP) being applied along with two reference risk levels that conceptually demarcate decision space into three regions.³² While the conceptualization and application of ALARP in the United Kingdom goes back to the 1940s, the origin of societal discussion on acceptable/tolerable risk in the nuclear field consists in the public hearings regarding the siting of Sizewell-B in the early 1980s.³³

In any case, judging whether a situation is “safe enough” or whether a certain risk is “acceptable” is intrinsically subjective. At the same time, safety goals involve an aspiration to gain “objective” validity when faced with the need to provide a societal justification on nuclear safety. In this context, the nuclear safety goals can be understood as a sociotechnical and ideal artifact to highlight the value-ladenness of safety decisions and an attempt to provide some reasonable accounts of these justifications. Thus, the safety goals straddle technoscience and society. Reflecting on these difficulties around the judgment of safety, the USNRC made a bold attempt to establish the safety goals through deliberations among experts and the public.³⁴ For example, the USNRC held a series of public meetings around the country from April to May in 1982 to receive public comments on the draft safety goals. In these meetings, the USNRC received 124 oral statements from various stakeholders, including industry, scholars, concerned citizens, professional societies, and others. Not all the interveners welcomed the proposed safety goals and their establishment process; the meeting record says “many individuals perceived the draft safety goals as ‘window dressing, an effort to assuage public fears, daily increasing, concerning accidents at nuclear power plants.’”³⁵ Although the manner in which the USNRC took into consideration those critical views in the following process lacked transparency, it is still worth mentioning that the USNRC dealt with the establishment of safety goals as a public issue and tried to make it transparent to put it on record.

Japan, too, has considered nuclear safety goals for almost 30 years in light of societal concerns. Toward the end of the 1980s, the Japanese nuclear industry, academia, and other related professional societies embarked on a consideration of safety goals in response to U.S. safety goals.³⁶ However, the Japanese government was slightly behind its overseas counterpart, with official studies initiated only in the 2000s.

Triggered by the reform of nuclear regulation shortly after the JCO Co., Ltd.’s Tokai nuclear fuel factory criticality accident in 1999, Japan conducted a substantial deliberation process to establish nuclear safety goals in the 2000s. The NSC of Japan, then an advisory body for nuclear safety regulation and policy, established a special committee comprising multidisciplinary members, including social scientists, representatives from labor unions, and consumer groups. The Interim Report of the committee proposed two types of safety goals, qualitative and quantitative, which were quite similar to those set by the United States in terms of their structures.³⁷ Following these draft safety goals, the committee suggested a set of draft performance objectives in January 2006, which were formulated as surrogates for the conformity assessments with safety goals to facilitate the risk management of reactor design and operation.^38,a

Despite the efforts made to consider safety goals and performance objectives, the NSC merely received the reports, but did not officially endorse them. The NSC did not clearly state their position on the significance of safety goals in the regulatory structure or their envisioned application in regulatory decision making. Sugawara and Inamura demonstrate that nuclear practitioners’ and experts’ fear of negative public perception and reaction when professing the existence of “mortality risk” has led to the NSC’s avoidance of explicit endorsement of safety goals.⁹ This nonfeasance continued until the Fukushima accident.

Drawing lessons from the Fukushima disaster, the newly established NRA commenced its study on safety goals immediately after its inauguration in September 2012. Following a half-year inquiry, a document entitled Discussion to Be Made Regarding the Safety Goals Before the Previous Meeting was authorized by the NRA, which reconfirmed the NSC’s Interim Report as a sufficient basis for discussion and suggested the need to include a new numerical goal, namely, the frequency of accidents, whereby the amount of discharge of ¹³⁷Cs exceeding 100 TBq at the time of the accident should not exceed 10⁻⁷/year.³⁹

The document makes important reflections on the lessons offered by the Fukushima accident, such as the consideration of long-term and wide-area land contamination. However, this “decision” by the NRA, as symbolized by the equivocal title of the document, avoided definitive statements about the safety goals and how these goals were to be implemented in the Japanese context. Furthermore, the NRA neither explained the basis for the decision nor made its deliberation process open to the public.

Surprisingly, the NRA has negated the nature of safety goals expected to act as a bridge between technoscience and society. An NRA official explained the NRA’s views on safety goals at a local public inquiry as follows: “The NRA has decided safety goals from a purely scientific and technical standpoint. It has nothing to do with social acceptance.”⁴⁰ A scholar criticized this approach as NRA’s regression to old-fashioned scientism,⁴¹ or “efforts to extend scientific ideas, methods, practices, and attitudes to matters of human social and political concern,” as coined by Olson.⁴²

In reaction to the “regulatory capture” criticism made after the accident,⁴³ the NRA has maintained a negative attitude toward communication with all external stakeholders. The underlying idea is that regulation will then be “purely scientific and technological,” and will not be influenced by any other factors. The following statement in the NRA’s Guiding Principles for Activities epitomizes the NRA’s view: “We shall make decisions independently, based solely on the latest scientific and technological information, free from any outside pressure or bias.”⁴⁴ The new “decision” of safety goals also reflects the regulator’s retreat to “science and technology” discourses. Although “how safe is safe enough” is a post-normal scientific question, not only requiring scientific inputs for its resolution but also a problem-solving activity different from the conventional notion of “science,”⁴⁵ the NRA seems to avoid possibly controversial value judgments by scientizing its framing.

III.C. Nuclear Emergency Preparedness and SPEEDI

The third case is that of the use of real-time simulation technology for nuclear emergency response.^8,b When the Fukushima nuclear accident occurred, SPEEDI, a real-time simulation system for radiation protection, failed in its mission. SPEEDI was officially expected to provide a fundamental scientific basis to undertake appropriate protection measures under a nuclear emergency based on the regulator’s emergency response guideline.⁴⁶ SPEEDI has a function to output charts that illustrate the contamination area and expected dose rate. This outcome is even more crucially important in a nuclear emergency in the Japanese context, considering the much bigger population and higher population density of Japanese nuclear-hosting areas than, for example, the typical situation in the United States. This characteristic of the siting environment of Japanese nuclear facilities requires setting necessary and sufficient protective actions and the zones in which such measures apply in order to make them practicable. However, such a chart was never released to assist the evacuation during the acute phase of the Fukushima disaster. SPEEDI became one of the focal points in the post-Fukushima dispute over off-site nuclear emergency management.

Evaluations on the usefulness of SPEEDI for emergency decisions were split evenly between two major investigation commissions: The National Diet of Japan Fukushima Nuclear Accident Independent Investigation Commission (NAIIC: the so-called Diet’s Commission) and the Investigation Committee on the Accident at the Fukushima Nuclear Power Stations (the so-called Cabinet’s Committee). These two major investigators showed almost opposite evaluations on the use of SPEEDI in the wake of the Fukushima nuclear disaster, as shown in Table II.

TABLE II Comparison of the Views of Two Major Official Investigation Commissions on the Fukushima Nuclear Accident*

	NAIIC	ICNAPS
General tone of the report	“Heated language, criticizing the decisions, behaviors, and general incapacity of those involved in the accident”	“Detached, objective narrative”
Evaluation on the use of SPEEDI in the wake of the Fukushima accident	“There were many intrinsic technical limitations to real-time prediction and advising against SPEEDI’s utilization.”	“SPEEDI was mishandled and squandered during the Fukushima accident.”
Views on SPEEDI, especially on its potential for nuclear emergency	Anti-SPEEDI: “We cannot use SPEEDI for future nuclear emergency as designated before Fukushima, due to its technical limitation mentioned above.”	Pro-SPEEDI: “Better procedures and more active use of SPEEDI to improve emergency response”

*By the authors based on Sugawara and Juraku.¹⁰ Quotes from Juraku.⁸

Consequently, the failure of SPEEDI has generally been framed as a problem of political secrecy (strong reluctance of the government to disclose inconvenient information) and technical insufficiency. Some social scientists have even critically analyzed this as a problem of structured secrecy.⁴⁷

However, as analyzed by us in another work,⁷ the fundamental failure of SPEEDI comprised neither secrecy nor the performance shortcomings of the computer system. The failure should rather be interpreted as a result of structural negligence of expertise on the intrinsic advantages and disadvantages of “atmospheric radionuclide dispersion simulation.”⁴⁸

One of the biggest and most fundamental limitations of SPEEDI is the difficulty of evaluating radioactive source terms for rapidly developing ongoing accidents. In principle, it is almost impossible to predict the time and amount of large releases of radionuclide from a damaged nuclear reactor. While one can roughly estimate the time at which the internal pressure of a primary containment vessel will reach the maximum design limit, it is not possible to predict the exact time or mode of rupture of a damaged reactor. That is, a damaged reactor could withstand even after the pressure increases, but this could result in a big rupture later. While leaks could begin from a seal or crack, a seal or crack could also prolong the leak, preventing us from seeing the explosive event. Thus, a damaged reactor is almost haphazard and is not at all predictable.

This problem is fundamental and not possible to be solved by the progress of our technical understanding of the system. It is due to our dependence on the behavior of the “complex system” that prevents us from making trustworthy predictions. Charles Perrow’s classical work on the theory of “normal accidents” made a similar point based on case studies of technological disasters, including the Three Mile Island nuclear accident,⁴⁹ and many other studies on the complexity in engineering and philosophy.

Thus, it would be difficult to utilize SPEEDI as a real-time automated consultation system that provides reliable predictions. However, SPEEDI should be used by decision makers with expert consultations under an appropriate advisory system to make their decision strategic, because SPEEDI would inform them of the variation of consequences of expected scenarios. It could also be used to share concrete images of a typical scenario of the radiological consequences of an accident with stakeholders before something unprecedented occurs at their neighboring nuclear facility. It might be a good idea to use SPEEDI calculation as an important reference to select the scenario of a nuclear emergency drill based on risk-communications among stakeholders. If we wish to reclaim the advantages of SPEEDI despite such limitations, it is important to establish a well-articulated institutional design of professional and technical consultation as other countries have attempted to accomplish.

All of these ideas regarding SPEEDI usage require deliberative institutional design to be implemented in advance. It is not promising to form it ad hoc in the wake of a disaster. A more fundamental and constructive perspective on the SPEEDI dispute has not been explicitly put forth in major investigation reports and has also been neglected by the newly established NRA. In 2016, the NRA denied the reliability of real-time radionuclide dispersion simulation and banned the use of SPEEDI and any other real-time radionuclide dispersion simulations for nuclear emergency decision making.⁴⁸ They now adopt the so-called concentric prescriptive action system instead of the adaptive real-time methods with the SPEEDI and real-time monitoring. The NRA has not detailed why it concluded that a real-time radionuclide dispersion simulation is totally useless for nuclear emergency response. In its policy document, it did not cite any investigation report or organize any public policy deliberation process on the use of such a technology.

Just five days before the NRA published its official statement, the Ministerial Council for Nuclear Power Utilization within the Japanese government released what could be read as a pro-SPEEDI recommendation, contradictory to the NRA’s policy.^c It maintained that the national government should not prevent prefectural or municipal governments from referring to SPEEDI calculation results on their own volition, either for nuclear emergency decision making or emergency drills.⁵⁰

As a result, all nuclear-hosting prefectural governments are required to make their own decision as to whether they will use real-time radionuclide dispersion simulation for nuclear emergency. As this is a difficult decision to make, many local government officers claim that it is impossible to use SPEEDI on their own initiative, without the formal endorsement of the national government. They also claim that it is difficult to legitimately make decisions and ensure accountability to citizens.^d

The NRA has already terminated the budget for the operation and maintenance of the SPEEDI computer system. The SPEEDI terminals have been removed from prefectural government offices. Effectively, hosting communities cannot use SPEEDI even if a nuclear accident occurs, despite the fact that the Ministerial Council’s judgment that SPEEDI could be used based on prefectural governments’ own volition remains active.

Similar systems to SPEEDI in other nuclear countries are neither banned nor naively relied upon in their nuclear emergency preparedness frameworks. According to our field work, France, Canada, and Sweden have their own strategic planning and institutional frameworks to take good advantage of the SPEEDI-like systems, with deliberate arrangements to utilize the expertise of the relevant professionals in a timely, effective, and legitimate manner. The authors do not attempt to evaluate their preparedness here, while we are aware that we should not apotheosize the other countries’ cases. However, it should be noted that they are all conscious of the importance of the expertise (and institutional arrangements, as an expression of the relevant expertise) and do not think that only the technical system (i.e., SPEEDI) is essential.

To sum up, a problem of the post-Fukushima controversy and the public policy process on nuclear emergency preparedness is that the deliberate and sensible use of relevant expertise has been structurally excluded. The emergency preparedness has shifted significantly from an emphatic pro-SPEEDI standpoint to an emphatic anti-SPEEDI standpoint. This is a journey from one extreme to another. The practical feasibility and technical rationality of the approach have been structurally neglected, and the problem remains unresolved.

IV. COMMON FEATURES OF JAPANESE “ACCEPTANCE” OF SAFETY/EMERGENCY CONTROVERSIES AND POLICY IMPLEMENTATION

By highlighting what has kept being ignored both before and after the Fukushima accident, this section extracts common features of Japanese “acceptance” of the safety and emergency controversies and policy implementation across the three cases.

IV.A. Avoiding Critical Conflicts

The essence of applying PRA and RIDM in nuclear safety may constitute tackling contradictory insights from deterministic and probabilistic approaches and striving toward integrating these approaches, and thus in providing a valuable opportunity for engaging in a serious debate on the fundamental conflicts of what nuclear safety is and how we ensure it. Establishing safety goals should have highlighted the socially constructive aspects of nuclear safety and opened the opportunity for stakeholders and the public to be proactively involved in discussions around safety and risk. The post-Fukushima controversy concerning SPEEDI may have focused on ways of making better use of relevant expertise, reflecting both the capability and limitations of simulation technology. These chances, however, have not taken advantage of even after the tragic accident.

What these three cases have in common is the lack of perspective toward a dynamic process for making nuclear safety more justifiable both technically and socially through the direct address of the conflicts between heterogeneous knowledge and contested values. This essential absence still prevails, regrettably. It may be said that the situation got worse if the post-accidental trajectory is understood as installing “right answers” instead of undertaking the responsibility of various conflicts, dilemmas, and trade-offs.

Although the NRA commissioners have not clearly stated this as such, regulatory avoidance is presumed to be relevant to the post-Fukushima social and political contexts. In such contexts, if the regulatory authority attempts to “relax” a part of the safety requirements in reference to the outputs from PRA and other considerations, the act will definitely be condemned as an act of compromise negatively influenced by “regulatory capture.” However, the situation in such contexts has not changed for the better, but has, in fact, deteriorated, with the public authority with oversight over nuclear safety prioritizing avoidance of criticism over the more pressing concern of enhancing public safety via profound discussions that address contested expertise and values.

The NRA’s circumventing of public deliberation on safety goals is also a case of avoidance of conflicts, risk trade-offs, and values. By arguing its peculiar “scientific” independence, the NRA successfully pretends that its goal is automatically derived from technical knowledge. Furthermore, the NRA does not permit any answer other than its own conclusion and uses its own performance target as the safety criterion to justify the sufficiency of safety upgrade measures for power station restart, despite the fact that this manner of using probabilistic risk criteria has often been cautioned as misleading and inappropriate.⁵¹ The NRA uses a small numerical number of the possibilities and consequences of severe accidents to settle social disputes that center on the public’s concern for the risk of Fukushima-class serious accidents.

In the case of SPEEDI, both the NRA’s ban on its use for nuclear emergency response and the Ministerial Council’s alternative judgment discussed practically nothing on the trade-offs concerning the uncertainty of real-time simulation technology. The NRA and Ministerial Council simply denied or admired the trustworthiness of SPEEDI output via a rough analysis, but never analyzed or discussed the multidimensional advantages and disadvantages of SPEEDI and their trade-offs on SPEEDI or concretely suggested alternative ways of using SPEEDI for emergency responses.

In these three cases, the core motivation for the relevant bodies’ behavior was the avoidance of critical conflicts among values, interests, and stakeholders. If these bodies could convince such conflicting actors by showing the “best scientific answer,” it should perhaps be the best way of averting—not solving—the social, economic, and political inconveniences caused by the risks of nuclear utilization. Considering that this could be represented not as a judgment made by someone responsible but as an absolute answer given by science, it alludes to the idea that no one can challenge such an answer and that it will always be defensible by the rhetoric, “it is the scientific fact.” This approach functions as an effective lubricant to carry out nuclear businesses without hassle.

By examining public controversies in Japan concerning the role of nuclear power and radioactive contamination, Polleri critically analyzes the METI’s way of framing nuclear power not as a policy option but as an apolitical necessity for the well-being of the Japanese nation-state and the common humanity.⁵² Both what Polleri coined as “post-political uncertainties” around the framing of nuclear energy policy and the avoidance of critical conflicts in the safety and emergency controversies elucidated in this paper has commonality in terms of evading public deliberation and scrutiny. While addressing and embracing conflicts should be essential both in the energy policy making in democratic societies and in RIDM in managing complex sociotechnical systems, the Japanese nuclear sector seems not to take responsibility for it.

Our analysis of the three cases in this section has shown the footprints of the Japanese nuclear safety construct. Here, we use the word construct as involving something that does not exist independent of, but relies on, the human mind for its existence.⁵³ Numerous social scientists construe safety as something constructed socially, culturally, and historically, such that social sciences can analyze, criticize, and intervene in its construction. From such standpoints, nuclear safety in Japan has tended to retain a historically shaped, path-dependent equilibrium among contested values. They have avoided critical conflicts, dilemmas, and trade-offs rather than embracing and resolving them. They have remained static, not dynamic, in contrast to the popular story.

IV.B. Proclivity Toward Automated Decision Making

As Martin Heidegger delineates in “das Man,” we, as human beings, are predestined to a limited existence.⁵⁴ Our knowledge of the universe is undoubtedly limited and the same is true of our knowledge of how the complex technical systems we make operate and interact with our society. This is the very reason why risk analysis and risk management decisions are made: to cope with such uncertainty. Nevertheless, as discussed previously, the Japanese ways of dealing with uncertainties are like a swinging pendulum: one pole comes to a halt in the face of uncertainties, while the other pole acts as though there were no uncertainties. It is clearly different from the other countries’ and international bodies’ narratives that attempt to legitimate their risk management by taking a specific balanced position between those two poles, as described in Sec. IV.A. Many critical analyses have argued that these narratives are not based on reality, and they have no negligible tendency to overstate their ability to control the risks. However, they still attempt to justify themselves by using this balancing rhetoric, while the Japanese perspective takes either of the poles in an outright manner. The background to such dichotomous behaviors is presumed to comprise a linear model, wherein the outcomes of a quantitative risk assessment are directly linked to management decisions and sociopolitical inclinations toward automated decision making. However, our analysis suggests that these underlying tendencies may hinder the effective application of expertise in making risk decisions in Japan.

In the pre-Fukushima era, the notion that “PRA is not useful due to its uncertainty” was widely adhered to by the nuclear sector before 2011, which justified the reluctance to apply PRA practices, PRA utilization, and RIDM in general.⁵⁵ Despite the large number of accident investigations and recommendations that suggest a wider and better use of PRA post 2011, the NRA repeatedly exaggerates the “incompleteness” and “immaturity” of PRA methodologies as an excuse for their hesitation in making risk-informed regulations.⁵⁶ We infer that such unchanged attitudes reflect a proclivity toward automated decision making, wherein the risk managers automate and scientize risk decisions to circumvent the responsibilities associated with judgments regarding how to embrace uncertainties, incertitude,⁵⁷ and our bounded rationality.⁵⁸ We cannot wait for the day when we understand everything about complex technoscience systems and assess their risk completely—that day will never come. That is why one strives to assess risks and uncertainties both quantitatively and qualitatively based on available information and adequate expertise as far as possible. On the contrary, the idea that PRA should not be taken into consideration unless its “completeness” is firmly demonstrated is similar to a conventional technocratic risk governance model, wherein straightforward comparisons are made between the numerical results of PRA and a set of safety goal surrogates. This can lead to “objective” binary decisions and to the automation of risk management. Such automation may negate the tenor of risk analysis and blur the roles and responsibilities of human judgment.

IV.C. Strategic Overlooking of Uncomfortable Knowledge

Confronting uncertainties and limitations in knowledge often requires risk managers to adopt adaptive risk management strategies, which constitute a dynamic, iterative, and structured process that includes regularly revisiting management objectives, monitoring and evaluating management outcomes, and incorporating new knowledge and lessons into future decisions.^59,60 Identifying high-profile issues and addressing them sensibly is key to the successful operation of adaptive management. However, our critical analysis reveals that this feedback process does not function properly in Japan, despite the fact that a number of nuclear professionals have recognized the issues that need to be challenged.

In the case of PRA, while this condition is epitomized by recognizing the need to expand the scope of risk assessment into various external events, there remains a delay in its implementation. Even long before the Fukushima disaster, some nuclear safety experts and risk assessors had recognized that the largest risk factors in Japan could be external events,⁹ although the PRA for external events was less experienced among Japanese experts and less developed than the PRA for internal events. They attempted to promote this by taking advantage of the Periodic Safety Review (PSR). Sugawara and Inamura revealed that in the 1990s, some key regulatory experts of nuclear safety had intended to make the industry expand its scope of risk assessment into external events and level 2 and 3 PRAs in the second round of PSRs after 10 years, while the limited scope of the first round of PSRs involved only internal events and level 1.5 PRAs were acceptable because the experts had placed importance on providing the risk profiles of each nuclear reactor at any rate.⁹ In reality, however, this aim was not well actualized. Except for seismic PRAs in accordance with the revision of the regulatory guide for seismic design in September 2006, the resources for expanding the assessment scope into other external events and level 2 and 3 PRAs had not been effectively allocated.⁶¹

While some attribute the delay in expanding the breadth of assessment to the exclusion of PRA from the regulatory scope when the PSR was made mandatory in 2002, the authors see it as a form of what Rayner called “uncomfortable knowledge,” or potential information that presents a sort of danger to institutions.⁶² As many academic studies and media coverage critically pointed out, Japanese nuclear power plants had been insisted as being “absolutely safe” before 2011 by almost all the nuclear proponents in this country.⁶³ Both PRAs and the draft safety goals had been mobilized to support such discourses. For instance, a nuclear safety expert explained that the calculated CDF and CFF of Japanese nuclear fleets were two or three orders of magnitude smaller than the proposed quantitative surrogates of safety goals, representing the considerable smallness of accident risks and sufficiently high level of safety. While this expert acknowledged that these numbers of CDFs and CFFs did not reflect the risks of external events, he thought the margin between the assessment results and proposed goals was enough to absorb them without no clear evidence.⁶⁴ If, by any chance, an extended PRA produces numbers that do not meet the proposed safety goals, the conventional ways of explaining safety would no longer be justified, and thus the existing edifice of the safety framework would be placed at immediate risk of failure. For the incumbent nuclear practitioners, enlarging risk assessment may produce uncomfortable knowledge that would jeopardize the safety discourse on which justification of nuclear energy use has heavily depended. This can explain why external event PRAs should remain, strategically or unquestioningly, as a mostly untouched issue; it may be listed as a future challenge, but has never seriously been considered.

Such uncomfortable knowledge can also be seen in the discourse of nuclear emergency preparedness. Even before the Fukushima accident, there had been a few opportunities for reviewing the prediction-based emergency arrangement. For instance, the Science Council of Japan insisted on the necessity of study in order to make evacuation decisions rapidly, reflecting the calculation time of Emergency Response Support System (a system expected to simulate accident progress and to provide source term information) and SPEEDI as well as the uncertainties of their results.⁶⁵ This argument, however, did not have enough political mobilization to change the basic concepts of emergency management. Moreover, after the Chuetsu-Oki earthquake hit the Kashiwazaki-Kariwa Nuclear Power Plant in 2007, the regulatory agencies and relevant ministries discussed the need for assuming a natural hazard-triggered technological event. Eventually, the discussion did not resulted in revisiting the existing emergency framework on the grounds that “it might fuel unrest and misunderstanding of local residents.”¹ There are also cases where uncomfortable knowledge that may endanger the extant institution is merely mentioned but strategically overlooked.⁶⁶

This “strategic overlooking” also applies to the case of safety goals. As the Interim Report repeatedly emphasized, the proposed safety goals were expected to begin with a process of trial uses to make further progress in risk assessment and management, involving the advancement of PRA models, the collection of a plant-specific datasets, and the accumulation of first-hand experience on RIDM (Ref. 37). In contrast to these original intentions, in practice, the draft safety goals had functioned as a tool for emphasizing an assertion that “nuclear power plants had already been safe enough.”⁹

The case of “societal risk” exemplifies the gap between original intention and reality. While in the U.S. safety goals “societal risks” refer only to total risks to life and health, in contrast to “individual risks,” the Interim Report attempts to capture the possible broader social impacts such as economic damage and land contamination caused by severe accidents. Since the available quantitative knowledge on the socio-economic consequences of nuclear disaster was quite limited at the time, the Report pointed out the need for further research in this area.³⁷ Unfortunately, little progress has been made in subsequent years. As Taniguchi notes, the fervor for severe accident research, including its impact on society, has witnessed a decline.⁷ We could find no content in relation to the social and economic consequences of severe accidents in the priority safety research plan formulated by the NSC during that time.

Dr. Shunsuke Kondo, Professor Emeritus of the Department of Nuclear Engineering at the University of Tokyo, speaking after the Fukushima disaster, expressed deep regret:

My biggest mistake was to select the probability of excessive exposure to ionizing radiation of the public as the surrogates of the safety goals. I was firmly convinced that safety meant protecting life. … I should have chosen the probability of occurrence of land contamination beyond the site boundary as the surrogate based on the level 3 PSA. It was a big blunder not to have learned any lesson pertaining to the seriousness of community collapse and further regarding the location of the off-site center from the Chernobyl accident.⁶⁷

Dr. Kondo, who had taken the initiative of introducing PRA and safety goals in Japan, reminds us of the severity and immorality of strategic overlooking.

The cases presented in this section show a common pattern; while enumerating on the important issues to be resolved in the initial phase of provisional institutionalization, these issues are, however, overlooked or neglected, whether deliberately or not, leading to the entrenchment of path-dependent structural problems.

V. STRUCTURAL IGNORANCE IN NUCLEAR SAFETY CONTROVERSIES AND POLICY PROCESS

The analysis above shows that the dynamism of safety has been ignored when addressing uncertainties, heterogenous knowledge, and incommensurable values can be a key for continuously reviewing the framework of nuclear safety. The authors here critically described how it remains structurally ignored despite the drastic change of main actors and political configuration due to post-accidental reform. This section further articulates what exists behind such persistence of Japanese society and thus conceptualizes it as structural ignorance in reference to the recent advancement of sociological studies of ignorance.

V.A. Deep Structural Roots of Ignorance

A pattern common to all the three cases is a reflection that the system and behavior of the Japanese nuclear program were inferior to, or lagged behind, “international standards,” thus resulting in a tendency to find an answer “out there” rather than figure out its own solution. As for PRA, as described in Sec. III.A, a new center was established within an industrial research institute under the auspices of METI to learn the “state-of-the-art” PRA methodologies. Symbolically, it was an American-based PRA expert who was appointed as the head of this new center. Moreover, while no clear evidence has yet been provided, the newly proposed goal of “100 TBq release” is said to be based on the numerical goals of Nordic countries. In the case of SPEEDI, the post-accidental reform was justified as introducing a “globally-established” emergency framework due to the recognition that no other countries adopted simulation-based emergency management.

Indeed, it is important and beneficial to refer properly to international common understanding and to learn about good practices in other countries. Japan has a long history of successes in importing products of advanced civilizations and institutionalizing them domestically; the import sources were China in ancient times; the United Kingdom, France, and Prussia at the dawn of modernization; and the United States after World War II. In the field of nuclear energy, the most prominent import country has been the United States, reflecting the historical contexts, namely, the ban of nuclear-related research by the General Headquarters and its removal by the San Francisco Peace Treaty, as well as a consistently strong political and economic bond with the United States in the era of postwar revitalization. Contrary to the “Not-Invented-Here Syndrome,”⁶⁷ technologies and systems invented in advanced Western countries have been regarded as being superior to those produced in Japan, and thus catching up with them and introducing them in the “right” way could be seen as a magic wand for overcoming accidents and disasters.

Such answers from the United States and international organizations, however, are not necessarily “established” or the “only” ways. For instance, PRA has been fiercely controversial from its beginning, as epitomized by a dispute around the WASH-1400 report.⁶⁸ Until gradually developing a concept of RIDM in which the PRA results would be prudently incorporated in safety decisions, the discussion around PRA has taken a detour, such as an ambitiously technocratic concept of risk-based decision making.⁶⁹ There are still a number of debates over how to implement the RIDM concept even in the United States, which has a long history of PRA and safety goals. To name a few, a proposal of fundamental change to a more risk-informed regulatory framework, including the adoption of a new “design-basis extension” category, was eventually disapproved by the NRC commissioners, while they expressed support for pursuing a risk-informed approach.⁷⁰ As for nuclear emergency management, there are some countries, such as Sweden, that try to take into greater consideration predictive simulation, contrary to Japanese reform, according to the authors’ interview with the person in charge of nuclear emergency preparedness at the Swedish regulatory agency.

In either case, there is no firmly established consensus. Rather, the safety-relevant concepts and methodologies such as PRA have been cultivated through a number of failures, misuses, and criticisms, in resonance with the respective societal and institutional environment in each country. When Japan tries to cite the “answers” developed in other societies, however, the contexts and controversies behind them are often discarded and cleansed. For Japanese nuclear proponents, having these imported goods becomes a new source of dispute that should be avoided because they want to introduce them to circumvent sociotechnical conundrums. While they also perfunctorily consider the historical context of PRA and safety goals, it is the finished products that they mainly focus on. In this way, deliberately or not, some aspects of these products that might possibly be a source of controversy are selectively ignored. Consequently, these products would be transformed into a materialistically and practically feasible form when institutionalized in Japanese society.

The ways of introducing risk governance serve as a typical case of such selective ignorance. The notion of risk governance was first mentioned in an advisory panel of the METI in 2013, which resulted in the frequent appearance of the words “risk governance” and the influential framework advocated by the International Risk Governance Council in many governmental documents of nuclear energy policy and industrial discourse. The concept of risk governance was originally developed when it was acknowledged that the conventional linear model, involving a clear distinction between scientific risk assessment and political decisions of risk management, was no longer valid when considering the uncertainties, complexities, and ambiguities of risk issues, as well as the conditionalities of risk knowledge.⁷¹ Reflecting on the ambiguity, and not the nonexistence, of knowledge regarding a possible mega-tsunami far beyond the design basis along the coastal area of Fukushima, and thus giving rise to manifold interpretations from different legitimate viewpoints, it should be rational to highlight the link between the ambiguity of risk knowledge and relevant organizational judgments within the risk governance framework. What subsequently transpired was, however, far from this original connotation; several nuclear utilities have attempted to establish company-wide risk management committees and to enhance public relations under the name of risk governance. The conceptual framework for acknowledging the pluralistic and multifaceted nature of risk issues has been simplified and reduced into institutionalizing in-house committees and strengthening old-fashioned explanatory activities.

Thus, by analyzing how the post-Fukushima change around nuclear safety and emergency management has been constructed in Japan, this study has elucidated the structural pattern wherein the Japanese nuclear program tries, deliberately or not, to selectively learn the concepts and methodologies of Western countries for neutralizing and cleansing uncomfortable knowledge, and thus introduce them as “answers” to the conundrum it faces.

V.B. Conceptualizing Structural Ignorance

Considering the deep structural roots of ignorance as described previously, the authors conceptualize it as structural ignorance in line with the recent development of studies on ignorance. What is referred to here as structural ignorance should be differentiated from other forms of ignorance. Structural ignorance is not the nonexistence or overlooking of knowledge pertaining to particular natural hazards or concrete accident scenarios. Rather, structural ignorance can be considered an extension of emerging strands of ignorance studies, particularly in the fields of sociology, and science and technology studies (STS). Some recent ignorance studies, often called agnotology, have elucidated that ignorance can be actively produced and intentionally mobilized when industries try to keep the question of the health consequences of their products open and exaggerate the absence of definite proof,^72,73 or when relevant governments and agencies try to circumvent blame and criticism in the aftermath of accidents or disasters.⁷⁴ Such perspectives and narratives of ignorance studies seemingly have a high affinity with those of the Fukushima nuclear accident in Japan. A number of its investigation works attributed the causes of this “man-made” disaster to deliberate ignorance of the possibility of huge tsunami and severe accidents produced by certain actors, including power industry associations and captured regulatory agencies, as found in the NAIIC investigation report’s narratives.^43,75

The post-Fukushima reform of nuclear safety governance was geared toward reducing the industry’s influence over policy making and preventing political capture of safety regulation. While the power of alleged “malicious” actors was significantly dampened, the post-accidental trajectory in Japan may demonstrate that ignorance has yet to be resolved, and worse, that ignorance has been preserved or even aggravated as described previously. This indicates that we cannot impute such ignorance to the strategic intentions of certain actors but to a more fundamental structural mechanism in Japanese society. This is the substance of structural ignorance.

The recent coming together of ignorance studies and STS has paid more attention to the institutional features of certain academic disciplines and scientific methodologies as well as to the norm underlying scientists’ and experts’ behavior.^76–78 Regarding nuclear safety, Downer sharply criticizes how not only the intentional and active production of uncertainties by the established organizational networks but also the very nature of formal reactor safety assessments, such as reliability calculation, systematically overstate their own completeness and certainty.^26,27 The authors share Downer’s perspective; however, their observation demonstrates that such a nature of the risk of relevant expertise is much more explicitly used to deny and forget uncertainties and incertitude of risks and our knowledge of them. It is not an opaque and implicit phenomenon but an honest and outright transformation in policy processes.

As analyzed in the previous sections, especially Sec. III, not just the nature of such expertise and the technocratic dream of nuclear promoters, but the different interests across stakeholders created a sociopolitical equilibrium that calls for an absolute and definite answer on various risk issues and the negligence of wisdom accumulated across the world. It is not created by the individual intentions and interests of stakeholders in Japanese nuclear governance but by the structure of their interrelations, policies, and decision-making processes and their interface with relevant experts and their expertise. It is persisting and repeating because the tacit and deep-rooted proclivity of the society has remained unchallenged.

Structural ignorance should also be connected with the concepts of “socio-technical imaginary,” “civic epistemology,” and “public reason” that have been suggested and elaborated by Sheila Jasanoff,^79–81 because civil society often played important roles in the mechanism of negligence of expertise as analyzed previously. People sometimes consider the technical tools and concepts for risk management as something “absolute,” “neutral,” or even “unchallengeable” than its reality—intrinsic underdeterminations. This epistemological belief and imaginary on the function and performance of the technology makes ill-equipped policy “reasonable” in public debate and decision-making processes. This is not a normative criticism of the “ignorance” of the public, but merely an observation made in order to determine the systematic and effective remedies to this structural problem.

To sum up, our analysis has articulated the structural nature of ignorance in a three-fold sense. At first, the dynamism in which undertaking heterogeneities, trade-offs, and conflicts would contribute to continuously revisiting the extant framework of safety has been structurally ignored even after the reform of governance and organizations. Second, there exists a historically repeated pattern of converting controversial conundrums to the issue of catching up “international standards” or introducing advanced products from abroad, instead of tackling them directly. Finally, when introducing the concepts and knowledge produced in foreign countries as being not necessarily “established,” the major actors in Japanese nuclear risk governance tend to selectively ignore their possibly controversial aspects. These deep-rooted mechanisms, which we would like to call the structural ignorance of expertise in policy process and its surrounding controversies, may contribute to conceal what led to the Fukushima disaster.

VI. CONCLUSION AND CHALLENGES: HOW CAN AND SHOULD SOCIAL SCIENCES CONTRIBUTE TO NUCLEAR SAFETY?

This study shows how the three key concepts and methodologies regarding nuclear safety and emergency preparedness have been interpreted, introduced, and/or abandoned before and after the Fukushima accident. It also identifies the following three common features: avoiding critical conflicts, proclivity for automated decision making, and strategic overlooking of uncomfortable knowledge. These features all involve ignorance of the dynamic nature of safety where addressing uncertainties, heterogeneous knowledge, and incommensurable values can be a key for continuously reviewing the existing edifice of safety. The authors articulated the deep roots of such ignorance and clarified the structural nature reflecting the societal and historical context. This can fittingly explain why ignorance persists in Japan despite the post-accidental drastic reform.

It may also be generalized as a common phenomenon of how countries and societies address nuclear risks. While this paper conceptualizes structural ignorance based on the Japanese cases, what aspects of safety are to be structurally ignored could differ depending on the historical and societal contexts in each country. In France, for instance, there exists some historical and sociological account of the ways of perceiving and dealing with nuclear accidents⁸² as well as of constructing counter-expertise,⁸³ which may be beneficial for examining the French version of structural ignorance and for further international comparative analysis. The authors find the possibility to extend the discussion on this point in connection with the concepts by Jasanoff as mentioned previously. She has analyzed similar policy process characteristics in science and technology from the point of view of “co-production.” It seems that she illustrates the mechanism as dynamic, evolutional, and creative. In contrast, the authors’ perspective focuses on static, resistant, and persistent sociotechnical and political structures in which such dynamism is spoiled, anchored, and stabilized. This connection could also help to find effective ways to break through the stubborn structure of ignorance for better policy processes and risk governance.

The authors believe that one of the roles of social sciences consists of resisting such ignorance while heightening the quality of public discourse around nuclear risks. While its concrete strategy needs further study, here we describe merely a possible direction: articulation of a volume of criticism and controversies experienced in the development process of technologies and concepts through “thick description.”⁸⁴ It can be particularly relevant to new emerging countries attempting to introduce nuclear energy. Theoretically and ideally, latecomers can take advantage of learning from first-movers without tracking the same difficult trajectories, which has accelerated the development of science and technology as well as human society itself. It is exactly the same reason why latecomers may ignore the contexts and controversies behind it. Ironically, formulating histories and lessons in a concise and understandable manner may facilitate the survival of structural ignorance. Rather, it is in detour trajectories that we can find profound lessons. For instance, Wellock⁸⁵ vividly delineates how the United States had taken the trouble to elaborate the ways of handling high-maintenance PRA tools amidst several conflicts and controversies, contrary to a simplified and glorified way of storytelling popular in Japan. The thick description of the history and the interplay of science, technology, and society, which some anthropologists, historians, and sociologists are good at producing, may help both the relevant actors in contemporary countries using nuclear power and the general public, particularly in emerging nuclear energy countries, to avoid the unquestioned acceptance of the conventional ways of constructing nuclear safety and to critically scrutinize what we tend to take for granted. Given the recent trends of the growing presence of new emerging countries in the field of nuclear power generation, the roles that the social sciences should play will only increase.

Acknowledgments

We offer our heartfelt thanks to the interviewees of our qualitative field work in France, Sweden, Canada, and Japan. Discussions during the Workshop of the Nuclear and Social Science Nexus: Challenges and Opportunities for Speaking Across the Disciplinary Divide Workshop, Organisation for Economic Co-operation and Development/Nuclear Energy Agency, in Paris, France, on December 2019, have been insightful for revising and extending this study. We also sincerely thank the three anonymous reviewers for critically reading the manuscript and suggesting substantial improvements. This work was supported partly by the JSPS KAKENHI grant number JP19K15271 and 17K18139. It was also supported partly by the Social Scientific Research Support Program on Local Community and Nuclear Power, Tokai village, Ibaraki, Japan, grant number TokaiRF201601.

Notes

^a The draft performance objectives in Japan proposed by the NSC consist of two indices: Core Damage Frequency (CDF) should be less than 10⁻⁴/year; and Containment Failure Frequency (CFF) should be less than 10⁻⁵/year.

^b This section is based on part of one of the author’s book chapters and has been extensively revised for this study (Juraku 2021).

^c The Ministerial Council system is an ad hoc and high-class policy coordination scheme preferred by the Abe administration to emphasize political initiative and serve as a response to past criticism of the strong and conservative Japanese bureaucracy.

^d This is based on interviews conducted in November 2016, January and March 2017, and October 2018 with prefectural and municipal government officials. The content is anonymized owing to requests made by the interviewees.

APPENDIX

LIST OF INTERVIEWS FOR THIS STUDY

The names of interviewees have been anonymized due to their consent.

A.I. INTERVIEWS IN JAPAN

The following interviews were conducted in Japan from September 30, 2016 to December 13, 2018:

September 30, 2016: Informant A, nuclear safety and emergency expert at private institution

September 30, 2016: Informant B, nuclear safety and emergency expert at public institution (retired)

October 14, 2016: Informant C, former national government officer in charge of nuclear emergency preparedness (retired)

October 30, 2016: Informant D, nuclear safety and emergency expert at public institution

November 17, 2016: Informants E and F, researcher of atmospheric dispersion at public institution

November 17, 2016: Informant G, municipal government officer in charge of nuclear emergency preparedness

December 27, 2016: Informant H, researcher of atmospheric dispersion at public institution

January 30, 2017: Informant I, officer at local nuclear hosting communities’ council

January 30, 2017: Informants J, K, L, M and N, prefectural government officers in charge of nuclear emergency preparedness (with informant H)

February 8, 2017: Informants O, P, and Q, researchers of atmospheric dispersion at private institution

August 2, 2017: Informant R, former national government officer in charge of nuclear safety regulation and research

October 19, 2018: Informant S, university professor of nuclear safety engineering (former prefectural government officer in charge of nuclear emergency preparedness)

October 19, 2018: Informant T, prefectural government officers in charge of nuclear emergency preparedness

October 20, 2018: Informants U and V, graduate student and professor of nuclear emergency preparedness at university

December 12, 2017: Informant W, researcher of nuclear safety in private institution

December 13, 2018: Informant X, university professor of nuclear safety engineering

A.II. INTERVIEWS IN FRANCE

The following interviews were conducted in France from November 21, 2016 to November 23, 2016:

November 21, 2016: Informants a, b, and c, officers of nuclear hosting communities’ council

November 22, 2016: Informants d and e, experts at TSO affiliated to the nuclear regulator

November 22, 2016: Informant f, academic researcher of nuclear emergency preparedness and post-accident recovery

November 23, 2016: Informants g and h, researchers of atmospheric dispersion at TSO affiliated to the nuclear regulator

November 23, 2016: Informant i, independent consultant of nuclear risk governance

A.III. INTERVIEWS IN CANADA

The following interviews were conducted in Canada from January 23 to January 26, 2018:

January 23 and 24, 2018: Group interview at the Canadian Nuclear Safety Commission

January 25, 2018: Informants j and k, professionals at nuclear operator

January 26, 2018: Informants l and m, university professors of nuclear safety engineering

A.IV. INTERVIEWS IN SWEDEN

The following interviews were conducted in Sweden on February 18 and 19, 2019:

February 18, 2019: Informant n, radiation protection expert at nuclear regulator (retired)

February 19, 2019: Informants o and p, university faculty members of philosophy of engineering

February 19, 2019: Informant q, officer in charge of nuclear emergency preparedness at nuclear regulator