CDM WORKING PAPER SERIES

Transcription

1 CDM WORKING PAPER SERIES Black Swan in Northeastern : Interdependent Systems, Escalating Disaster on March 11, 2011 Aya Okada and Louise K. Comfort Working Paper: 1105 November CENTER FOR DISASTER MANAGEMENT Graduate School of Public and International Affairs University of Pittsburgh 3913 Wesley W. Posvar Hall 230 S. Bouquet St Pittsburgh, PA March, 2011 The views expressed herein are those of the authors and do not necessarily reflect the views of the Center for Disaster Management and the University of Pittsburgh. University of Pittsburgh, By Aya Okada and Louise K. Comfort. All rights reserved. Working Paper: Not for citation without consent of the authors 1

2 Black Swan in Northeastern : Interdependent Systems, Escalating Disaster on March 11, 2011 Aya Okada and Louise K. Comfort CDM Working Paper No APPAM Paper No.: PAPER2640 November 2011 Abstract To what extent and how should investments be made for Black Swan events, that rarely happen but trigger extreme consequences? The extraordinary disaster that struck on March 11, 2011 an earthquake with magnitude 9.0, series of tsunami that reached as high as 40.5 meters inundating 561 square kilometers of land, and a nuclear reactor breach at Fukushima Daiichi provides test case for examining this policy question. The paper reviews laws and plans for disaster management in to examine the extent to which the country had prepared for prior to March 11 th, and contrast this with observed practices, focusing on interactions and communication flows among organizations engaged in responding to the disaster. Data on practices were collected through content analysis of newspaper articles and field observations. The analyses revealed three distinct networks of responding organizations: earthquake, tsunami, and nuclear threat. We observe significant difference in composition and structure of these networks, particularly between the first two and the network for nuclear threat. We also reveal an increase in efficiency of communication flow when networks responding to the three policy areas are brought together. We conclude with a recommendation to create knowledge commons across multiple policy areas in planning for rare and extreme events. Aya Okada Graduate School for Public and International Affairs Center for Disaster Management University of Pittsburgh Pittsburgh, PA ayo2@pitt.edu Louise K. Comfort Graduate School for Public and International Affairs Center for Disaster Management University of Pittsburgh Pittsburgh, PA lkc@pitt.edu 2

3 1. Introduction: Planning for Rare and Extreme Events Planning for events that rarely happen presents a major challenge for policy makers. Difficult yet important decisions must be made with regard to allocation of scarce resources and time. In facing the trade-off between preparing for rare events and managing routine needs, policy makers must ask the following two questions: to what extent and how should investments be made in preparing for rare events? Policy makers cannot overlook events that rarely occur, as failure to do so increases risk of extreme consequences when such an event does happen. The Great East Earthquake (or 2011 Tohoku Earthquake and Tsunami) that struck on March 11, 2011 is a perfect example of such a rare and extreme event. An earthquake with magnitude of 9.0, the largest earthquake ever recorded in, occurred off the cost of Miyagi Prefecture in Northeastern. The earthquake triggered a tsunami that reached as high as 40.5 meters (132.9 feet), flooding 561 square-kilometers of the land (Cabinet Office, 2011a). The tsunami damaged nuclear reactors at the Fukushima Daiichi Nuclear Power Plant, destroying its cooling system. This led to a nuclear threat which International Atomic Energy Agency (IAEA) evaluated as the highest Level 7 in International Nuclear Event Scale. With the release of radioactive materials, more than 113,000 residents were (and still are) restricted from returning home 1. The Great East Earthquake is exactly what Taleb (2007) called a Black Swan event. Taleb gives three distinct characteristics for such events: 1) the event comes as a surprise to the observer, 2) the event has major impact, and 3) the event is rationalized only by hindsight. In addition to the magnitude of the earthquake and tsunami, relevant ministries, agencies, and even researchers had not predicted three complex, interacting events to occur with near simultaneity. The massive earthquake triggered a continuous rupture of a fault line stretching for a length of 450km (279 miles) with a width of 200km (124 miles) (Cabinet Office, 2011a). The earthquake, in turn, generated a severe tsunami, which flooded the Fukushima Daiichi Nuclear Plant, resulting in the failure of the back-up generators and releasing damaging radioactive materials into the environment. While was well prepared for the earthquake, neither the emergency management community nor residents of the areas were ready for a tsunami of such 1 A figure given by Tadahiro Matsushita, Vice-Minister of Economy, Trade and Industry at the General Affairs Committee, House of Representatives, on June 16,

4 large size to occur. Nor was Tokyo Electric Power Company (known as TEPCO), a private firm running the Fukushima Daiichi Nuclear Power Plant, prepared for such an extraordinarily damaging event. The unexpected sequence of events led to an impact that was also beyond one s imagination. 15,828 people died, 5,942 people were injured, and 3,760 people are still missing as of October 21, 2011 (NPA, 2011). The entire economic impact has been estimated to reach as high as 25 trillion yen or US$ billion 2 (Cabinet Office, 2011a), and the government continues its struggle today to secure sufficient budget to cover all the damage and losses. Needless to say, people in are still struggling to find a way to rationalize what has been termed as unprecedented disaster ( 未曾有の大震災 ), washing away towns and villages in Northeastern. In addition to the surprising scope and magnitude of the impact, the Great East Earthquake on March 11 illustrates interdependencies among large-scale sociotechnical systems that serve basic functions for continuity of operations in advanced industrialized societies. has recognized the need to respond to earthquakes that continue to occur hundreds of times all across the country, and to respond to the damage and the risk of further tsunamis. The nuclear threat added further complications, leaving people without knowledge of nuclear power plants and radioactive materials at a complete loss. Even the government agencies and experts, who had not assumed a crisis of such severity, did not know the appropriate steps to take. With the risk of further earthquakes and tsunamis, several power plants stopped their operations, and half the country faced the risk of losing electricity, even the capital, Tokyo. With these crises occurring virtually at the same time, actors with multiple perspectives were engaging in the same public arena, bringing different types of expertise to address the consequences. They faced the challenge of communicating with each other, often using terms that were not mutually understood, and making timely decisions for actions to protect the affected communities and address the long term needs of the society. Taking the Great East Earthquake as a case of rare and extreme events for which policy makers must prepare, we argue that it is essential to recognize the interdependencies among different types of disasters and the consequent need to adjust and integrate planning for response operations under uncertainty. Such planning and adaptation can be achieved through creating a knowledge commons (Hess and Ostrom, 2006) to facilitate efficient communication 2 Calculated with an exchange rate of US$1=76 yen. 4

5 and effective decision-making among actors that are responsible for managing these catastrophic events. Following the theoretical framework of this argument, the paper reviews the preparations that had made prior to the Great East Earthquake on March 11, 2011, primarily focusing on disaster management laws and plans. We then present analyses of the actual response to the event based on results from network analyses and field observations. We discuss discrepancies observed between plans and practice, and the inadequacy of preparation in disaster management. We conclude with the need to analyze the interdependencies among natural hazards and human response, and the recommendation to create a knowledge commons across multiple policy areas in planning for rare and extreme events. 2. Managing Large-scale Sociotechnical Systems Three streams of literature are relevant to an analysis of extraordinarily catastrophic events such as the March 11, 2011 disaster in Northeastern. First, the fundamental challenge of building sustainable interactions between coupled human and natural systems is underscored by both the scope and the difficulty of these three interdependent events. This challenge, framed by Charles Perrow (2006) as a pattern of reciprocal interdependencies between large scale technical infrastructures and the communities that design, build, and benefit from them, is an inescapable conundrum for public policy makers. In essence, the very technical infrastructures that we build to support and advance normal operations in human communities also retain the capacity to damage these communities severely if they fail. Further, designed by humans with limited cognitive capacity, these same technical structures are vulnerable to failure under unforeseen and unrecognized risks. A potential counter to Perrow s sobering argument is the advance of information technology to extend human cognitive capacity on a community-wide scale. Nobre, Tobias, and Walker (2009) argue that innovative uses of information technology can facilitate collective learning and enable communities to recognize risk more accurately and to mobilize collective action more efficiently in reducing the likely impact of severely damaging events. Such an approach places the responsibility for informed action across the spectrum of jurisdictions, organizations, and disciplines. Collective action, consequently, is enabled by a steady, conscious 5

6 effort to understand the environment in which one lives and a systematic approach to engage multiple, diverse actors in assessing the risks to which the community is exposed, identifying the resources available and strategies of action that are possible to reduce those risks, should the damaging event occur. This approach emphasizes the potential for information technology to strengthen human reasoning capacity, rather than replace it, as it assumes that the changing conditions require a continuous ability to assess, update, and reframe strategies of action to adapt effectively. A third approach that lends perceptive insight to this complex problem is that of collaborative rationality introduced by J. Innes and D. Booher (2010). The authors argue that complex problems that necessarily involve interactions among multiple stakeholders with diverse interests require a different form of problem solving to reach a sustainable resolution of the problem. The authors assert that standard policy processes assume a largely static environment and actors unable or unwilling to learn. They assert that solutions for dynamic environments must necessarily create a public forum for the open exchange of information, regular monitoring and assessment of risk, and continuous learning by the majority of actors involved in addressing the problem. The strength of this approach is that, in practice, collaborative rationality builds community commitment to manage risk. Its weakness, however, is that some actors may choose not to learn, not to participate, not to reduce risk to the community, but instead to advance their particular interests at the expense of the community, thus exposing other members of the community (and themselves) to even greater risk. The three approaches share a common recognition of the roles of individual and organizational learning to reducing risk. While the specific methods for doing so are not yet well developed nor applied in different contexts, they offer a constructive set of insights into the potential development of a knowledge commons to reduce disaster risk. Such a knowledge commons, based on Elinor Ostrom s initial concept and further augmented in her work with Charlotte Hess, is characterized by three general characteristics. First, the community creates a public space for sharing information, whether it is in the town square, as in medieval times or more recently in virtual commons as shown in the Arab Spring in Second, the community creates a technical infrastructure for the rapid and easy exchange of information among the participants, an infrastructure that all participants may access, whether they choose to do so or not. Third, an emergent leader or group successfully articulates the primary goals of the 6

7 community in language and terms that everyone can understand, and which form the basic call to action. While these characteristics are likely essential for developing a knowledge commons, they are not necessarily sufficient. The drive to build the commons will likely come, rather, from the dynamic interactions of the groups most affected by the risk. 3. Disaster Management Prior to March 11, 2011, a small island country at the tip of Asian continent is a disaster-prone country with high risks of earthquakes, tsunamis, volcanic eruptions, and typhoons. Located on the Pacific Ring of Fire, has experienced 20.8 percent of the large-scale earthquakes (magnitude 6.0 or higher) around the world (Cabinet Office, 2011b). Well-known is the Kobe Earthquake of 1995 (formally known as the Great Hanshin-Awaji Earthquake), which resulted in more than 6,400 casualties. The country is also a home to 108 active volcanoes. Given its high risk of natural disasters, as well as man-made disasters, has invested substantially in creating a strong disaster management system since the 1940s. Between 1995 and 2004, an average of 4.5 trillion yen or US$59.2 billion 3, has been designated for disaster management. This accounts for approximately five percent of the national general account budget (Cabinet Office, 2011b). The current disaster management system is built upon the Disaster Countermeasures Basic Act, first enacted in 1961 after the Isewan Typhoon in The Act: 1) clarifies responsibilities of relevant organizations and citizens in disaster management; 2) designates organizations specific to disaster management; 3) designates plans and systems for disaster management; 4) defines roles and responsibilities in three stages of disaster management (preparedness, response, and recovery); 5) addresses financing for disaster management; and 6) states definitions and procedures for declaring states of disaster emergency. Based on this Disaster Countermeasures Basic Act of 1961, operates the disaster management system as described in Figure 1. 3 Calculated with an exchange rate of US$1=76 yen. 7

8 Figure 1: Outline of the Disaster Management System Source: Cabinet Office (2011b). While the main responsibility for preparedness and response is granted to both national and local governments (prefectural and municipal), the Cabinet Office plays the main role in organizing the management of disaster preparedness, response, and recovery. Its role has been enhanced from lessons learned from the Kobe earthquake in For example, the Cabinet Information Research Center was established within the Cabinet Intelligence and Research Office in 1996, and the post of Minister of State for Disaster Management was established under the Cabinet Office in The Minister leads the disaster management team in the Cabinet Office and acts a member of the Central Disaster Management Council. Based on the Disaster Countermeasures Basic Act, the Central Disaster Management Council 4 drafts and promotes the Basic Disaster Management Plan, the action plan for practices in disaster management in. The Basic Disaster Management Plan addresses comprehensive and long-term disaster reduction strategies. These include disaster management related systems, disaster reduction projects, early and appropriate disaster recovery and rehabilitation, as well as scientific and technical research. 4 The Council is chaired by the Prime Minister, and the members include all cabinet members (including the Minister for State for Disaster Management), heads of four designated public corporations (Governor of the Bank of, President of ese Red Cross Society, President of Broadcasting Corporation or NHK, and the President of Nippon Telegraph and Telephone Corporation), and four experts. As of September 12, 2011, the four experts are: professor in earthquake countermeasures, professor on risk management, a governor representing Association of Prefectural Governors, and board member from Fire Association known as Nissho (Central Disaster Management Council, 2011a). 8

9 3-1. Types of Disasters Addressed in the Act and the Plan The Disaster Countermeasures Basic Act as well as the Basic Disaster Management Plan focuses on 13 different kinds of disasters that may encounter (Table 1). This section reviews major plans that had prepared for earthquake, tsunami, and nuclear disasters prior to the Great East Earthquake on March 11, Table 1: Disasters in Basic Disaster Management Plan Natural Disasters ( 自然災害 ) Accident Disasters ( 事故災害 ) Maritime Disaster Aviation Disaster Earthquake Disaster Railroad Disaster Storm and Flood Road Disaster Volcano Disaster Nuclear Disaster Snow Disaster Hazardous Materials Disaster Large-scale Fire Disaster Forest Fire Disaster Source: Basic Disaster Management Plan (2008) Earthquakes Countermeasures for earthquakes consisted of enhancement of observation systems, identification of potential large-scale earthquakes, enhancement of resistance in buildings, and preparation for tsunamis. Meteorological Agency (JMA) leads the observation of seismic activity around, with 4,119 observation points all across the country (Cabinet Office, 2011b). In 2006, the Earthquake Early Warning (EEW) Information system was established, where P-waves detected near the epicenter is immediately processed to alert the citizens before the S-wave arrives. The system has been made available to the public since 2007, and today in many instances, the general public receives the early warning for potential earthquakes on television and radio. Predicting massive earthquakes have been a major concern for the ese government. The Plan has identified four potential earthquakes with high possibilities of occurrence in the coming years: Tokai Earthquake, Tonankai and Nankai Earthquakes, Earthquakes around the and Chishima Trenches, and Tokyo Inland Earthquakes. To better prepare for such 9

10 earthquakes and to mitigate the damage, a Policy Framework (a master plan), Earthquake Disaster Reduction Strategy, and Guidelines for Emergency Response Activities have been established for each of the four potential earthquakes. Figure 2: Predicted Scale and Damage of Four Potential Earthquakes Source: Cabinet Office (2011b) Tsunami Although has had some experience of tsunamis (Table 2), the government had not identified tsunami as a disaster category that merits attention on its own, and has treated countermeasures for tsunamis as a need generated by earthquakes. The countermeasures include enhancement of warning system, improvements of coastal embankments (tidal embankments) and tide prevention gates, as well as enhancement on tsunami hazard maps. 10

11 Table 2: History of Tsunami in Name Year Number of Casualties and Missing People Meiji Sanriku Earthquake Tsunami 1896 Approximately 22,000 Showa Sanriku Earthquake Tsunami ,064 Tonankai Earthquake ,223 Nankai Earthquake ,443 Chile Earthquake Tsunami Tokachi-oki Earthquake Nihon-kai-Chubu Earthquake Hokkaido-Nansei-oki Earthquake Source: Cabinet Office (2011), p.32. While no national plan has been made specifically for tsunamis, regions with past experiences of tsunami have made investments for such potential risk. Taro district in Miyako City, Iwate Prefecture is one such example. This district has experienced two tsunamis during the past 120 years: 15m-high Meiji Sanriku Earthquake Tsunami in 1896, and 10m-high Showa Sanriku Earthquake Tsunami in These experiences have led the community to declare to be the town for tsunami disaster management and to construct three 10m-high sea walls that stretch for a total of 2.4 km (1.78 miles). The district also avoids right-angle corners in crossings to facilitate faster evacuation in case a tsunami approaches. Figure 3. Tsunami Countermeasures in Taro District, Miyako City Two signs at the height of past tsunamis Sea walls known as The Great Wall of China Source: Pictures taken by the authors on November 1,

12 For all natural disasters including earthquakes and tsunamis, the ese Governments operates the Integrated Disaster Management Information System to assess the disaster situation early in the incident and promote information sharing among relevant organizations, thereby enabling quick and appropriate decision-making for emergency response operations (Cabinet Office, 2011b, p.16). The main features include: 1) Early assessment of damage from earthquakes: The System receives information on earthquake intensity as observed by the JMA and automatically activated by an earthquake intensity level of 4 or greater. This function estimates the distribution of seismic intensity and scale of damage (human suffering and building damage) within 10 minutes. 2) Early damage assessment using satellite observation systems: This function uses images taken from artificial satellites capable of wide-area observation to provide early assessment of damage. 3) Information sharing: This function plots disaster incident locations by disastermanagement agencies to a map using GIS, so it can be freely accessed by all. For sharing information on disaster response, various agencies have developed their own Disaster Information Sharing Platform. Studies have analyzed the way to integrate different systems and interfaces (e.g. Kusakabe et al. 2007), but such systems remain fragmented in municipalities and prefectures, although efforts have been made to specify standards for such information platforms (Kanawaza et al. 2007) Nuclear Disasters Nuclear disasters are identified as one types of accident to which is exposed to in the Basic Disaster Countermeasures Plan., as of December 2009, had 17 nuclear power plant stations, with 54 reactors operating. Nuclear power had accounted for 13.4 percent of the total electricity generated in the country, with an increase to 23.2% expected by 2030 (AEC, 2009). The 17 nuclear power plants are operated by private electricity firms 5, termed as the nuclear operators in the Act and the Plan. 5 There are 10 such firms: Atomic Power Company, Hokkaido Electric Power Company, Tohoku Electric Power Company, Tokyo Electric Power Company, Chubu Electric Power Company, Hokuriku Electric Power 12

13 Until the nuclear threats triggered by the Great East Earthquake at Fukushima Daiichi Plant in March, 2011, had experienced nine major nuclear accidents during the 1990s and 2000s (Table 3). Among them only one was caused by an earthquake: the accident at Kashiwazaki Kariya Power Plant in While the incident did not cause fire or a release of radioactive materials, Hamaoka Nuclear Power Station in 2009 also stopped its operation immediately after an earthquake off the cost of Shizuoka Prefecture (magnitude 6.5) in Year Table 3: Major Nuclear Accidents in during the 1990s and 2000s Name of the Power Plant 1991 Mihama Power Plant 1993 Tokai Reprocessing Plant 1995 Monju Nuclear Power Plant 1997 Tokai Reprocessing Plant 1999 JCO Uranium Reprocessing Plant Operating Company Accident Rates in International Nuclear Event Scale Kansai Electric Power Company Power Reactor and Nuclear Fuel Development Corporation (Donen) Power Reactor and Nuclear Fuel Development Corporation (Donen) Power Reactor and Nuclear Fuel Development Corporation (Donen) Nuclear Fuel Conversion Co. (JCO) 1999 Shiga Power Plant Hokuriku Electric Power Company 2004 Mihama Power Plant 2007 Kashiwazaki Kariya Power Plant Kansai Electric Power Company Tokyo Electric Power Company 2008 SHIEI S.H.I.Examination & Inspection,Ltd. A break in a tube of a steam generator led to full activation of Emergency Core Cooling System. Small amount of radiation released to the outside A worker exposed to radiation while conducting regular check-ups.. The heat was so intense that it warped several steel structures in the room. [3] A Fire explosion at a uranium fuel reprocessing site. At least 37 workers were exposed to inhale doses of radiation. Critical accident in a uranium reprocessing facility. Two workers were killed and several hundreds of residents were exposed to radiation. Mishandle of a system to control water pressure resulted in part of the reactor core turning critical. A break in a pipe caused high temperature steam to leak. Four workers were killed, and seven others were injured. Fire breakout after the Chuetsu Offshore Earthquake (magnitude 6.8) triggering a release of some radioactive material. A worker exposed to radiation while repairing cyclotron facilities. 2 (Incident) 2 (Incident) 1 (Anomaly) 3 (Serious Incident) 4 (Accident with Local Consequenc es) 2 (Incident) 1 (Anomaly) 0 (Deviation) 2 (Incident) Source: Seibido Mook (2011), pp Company, Kansai Electric Power Company, Chugoku Electric Power Company, Shikoku Electric Power Company, and Kyushu Electric Power Company. 13

14 Safety measures for nuclear disasters in consist of three pillars, with the purpose of managing exposure to radiation, prevention of accidents, and mitigation of the effects of the accidents (AEC, 2009, p.43): 1) Reducing the amount of radioactive materials released during normal operation 2) Employing multifaceted protective mechanisms 3) Enhancing preparedness measures Structure of organizations involved in dealing with nuclear disasters is quite complex, as is illustrated in Figure 3 describing an example of emergency response system. Agencies are spread across multiple levels of governance, and involved organizations include the Nuclear and Industrial Safety Agency (NISA) under the Ministry of Economy, Trade and Industry (METI) as well as the Nuclear Safety Commission (NSC) under the Cabinet Office. There are multiple quasi-governmental organizations involved, often known as the independent administrative agency, as is seen in an example of Atomic Energy Agency (JAEA). Figure 4: Nuclear Disaster Management System in Case of an Emergency Source: Atomic Energy Relations Organization (2009) as cited in AEC (2009). 14

15 More detailed responsibilities for safety measures and response operations in case of disaster are stated in the Act on Special Measures Concerning Nuclear Emergency Preparedness, enacted in According to this Act, nuclear operators are obliged to report to the competent minister in case of a detection of potential nuclear disaster; upon such report, the Prime Minister has the authority to announce A Declaration of Nuclear Emergency Situation. In such an event, the Act grants the Prime Minister emergency power to give orders not only to relevant agencies in national and local governments, but also to private firms operating the nuclear power plants. A brief overview of the laws and plans for disaster management in reveals three important characteristics that affected the response system to the Great East Earthquake on March 11, First, while elaborated in the same Disaster Countermeasures Basic Act and the Basic Disaster Management Plan, different types of disasters were treated separately with distinct measures for preparedness, response, and recovery. This was true not only for the basic preparedness plan, but also for assumed response and recovery structures. Almost no reference is made to the possibility of simultaneous occurrence of multiple types of disasters. Secondly, since tsunamis were considered to be generated by earthquakes, less attention and planning for countermeasures were given to this threat. Such an attitude for tsunami is reflected in an incident where TEPCO did not report the results of the simulation to central government. TEPCO had found, three years ago, that based on data from Meiji Sanriku tsunami of 1896, Fukushima Daiichi Nuclear Power Plant risked facing a tsunami of 15 meters (49.2 feet). However, the estimate and the need for enhancing tsunami countermeasures were not reported to the Nuclear and Industrial Safety Agency (NISA) until four days before the Great East Earthquake (Yomiuri Newspaper, August 25, 2011). Thirdly, public organizations, especially ministries and agencies in the central government, are viewed as primarily responsible for managing disaster preparedness, response, and recovery. The Act and the Plan assume public organizations to be equipped with the capacity to undertake the responsibilities in dealing with disasters. 15

16 4. The Great East Earthquake, March 11, 2011 The Great East Earthquake far exceeded s laws and plans for disaster preparedness. First, the magnitude of the earthquake was far greater than predicted. Although the Trench-type Earthquakes in the Vicinity of the and Chishima Trenches were part of it, the seismic rupture had an extraordinary of length of 450km (279 miles) with a width of 200km (124 miles) (Cabinet Office, 2011a). Researchers in the Seismological Society of bravely stated that the Great East Earthquake was a defeat of seismology in (Nikkei, October 15, 2011). Second, an earthquake of such unpredictable scale triggered a massive tsunami, which had not occurred in the area for more than fifty years after the Chile Earthquake Tsunami in Reports reveal that while residents in Tohoku area had successfully evacuated to designated shelters after the earthquake, many people lost their lives for not believing that a tsunami could reach where they were (Central Disaster Management Council, 2011b). Finally, tsunami was not considered to be an important risk factor for nuclear power plants; it was even imagined that an incident that engages Prime Minister s emergency declaration would never occur. Given the inadequacy of preparedness for disasters that occurred on March 11, 2011, how did organizations designated in the disaster plans and in the affected communities respond in actual practice? In this section, we present a descriptive analysis of response systems that emerged during the three weeks immediately after March 11, The response system was identified through a content analysis of newspaper articles from Yomiuri Newspaper 6, from March 11 to April 1, 2011 (627 articles). We identified organizations that made an action as a responding organization, and coded their funding sources (public, private, and nonprofit) as well as jurisdiction (international, regional, national, prefectural, municipal, local 7 ). Actions that involved interactions and/or communication with other organization(s) were coded, and these measures were used for network visualization and calculation of centrality measures. The network analysis was calculated using ORA, a software program for network analysis developed at the Center for Computational Analysis of Social and Organizational Systems (CASOS), Carnegie Mellon University. We complement the results obtained from network analyses with 6 The Yomiuri Newspaper is one of the five national newspapers in with a daily circulation of approximately10 million (Yomiuri Newspaper, as of July, 2010). 7 Local organizations are those located in the three affected prefectures: Iwate, Miyagi, Fukushima. 16

17 field observations made between October 29 to November 2 nd, 2011 in Miyagi and Iwate Prefectures as well as data obtained from interviews in Tokyo Network Analyses The content analysis of newspaper articles identified 1,101 organizations involved in the response network. This analysis captures only those organizations documented in news reports, and does not claim to represent the entire response system. Given the broad area affected by the disaster, and the ripple effects that followed, it is difficult to find organizations in that did not take some action in response to the disaster. Nonetheless, the networks identified from the content analysis provide a general profile of the organizations that were involved in responding to the series of disasters. As Table 4 and 5 show, 1,101 responding organizations varied by funding source as well as jurisdiction. As was implied in the Disaster Countermeasures Basic Act and the Basic Disaster Management Plan, public organizations accounted for 53.6% of the responding organizations; similarly, almost half of the organizations were national (47.0%). Private organizations accounted for 35.9% of the total number of organizations, and approximately 15% of both international and local organizations. Table 4. Organizations by Jurisdiction and Funding Source (N=1,101) Jurisdiction Number of Funding Number of Organizations Organizations (%) Source (% in each jurisdiction) Public 122 (73.5%) International 166 (15.1%) Private 36 (21.7%) Nonprofit 8 (4.8%) Public 4 (12.5%) Regional 32 (2.9%) Private 27 (84.4%) Nonprofit 1 (3.1%) Public 215 (41.5%) National 518 (47.0%) Private 234 (45.2%) Nonprofit 69 (13.3%) Public 91 (65.0%) Prefectural 140 (12.7%) Private 33 (23.6%) Nonprofit 16 (11.4%) Public 51 (67.1%) Municipal 76 (6.9%) Private 18 (23.7%) Nonprofit 7 (9.2%) 17

18 Local 169 (15.3%) Public 107 (63.3%) Private 48 (28.4%) Nonprofit 14 (8.3%) Key organizations identified in ORA (Figure 4) 8 also revealed the same trend, with eight out of 10 organizations being public and national. The exceptions were Tokyo Electric Power Company (private/national) and Saitama Prefecture (public/prefectural). Political parties appeared as important organizations for their role in initiating discussion on securing budgets for relief and recovery during the first three weeks. Figure 5: Key Entities in the Overall Response Network Legend: jp_pm: Prime Minister, dpj_sg: Office of Secretary General, Democratic Party of tepco: Tokyo Electric Power Company (TEPCO) japan_gov: ldp: Liberal Democratic Party pnp_prc: Chairman of Policy Research Council, People's New Party saitama: Prefectural Saitama cab_vmin: Vice Minister, Cabinet Office, dpj_diet: Chair for Diet Affairs Committee, Democratic Party of ldp_diet: Chairman, Diet Affairs Committee, Liberal Democratic Party 8 ORA identifies key entities within the network by calculating a set of centrality measures: total degree centrality (controlling agency), Eigenvector centrality (inter-organization leader), hub centrality (acts as a hub), authority centrality (acts as an authority), betweenness centrality (information conduit), and high betweenness and low degree centrality (connects agencies). Key entities are thus organizations with the most ability to monitor and direct information flows and to reach out to other organizations to perform tasks and gain resources. Refer to footnote #10 for basic explanation of four key centrality measures. 18

19 Table 5: Top 10 Organizations in Four Centrality Measures 9 Degree Centrality Betweenness Centrality Closeness Centrality Eigenvector Centrality Prefectural Prime Minister, Minister of Foreign Affairs, Saitama Office of Secretary General, Democratic Party of Prime Minister, Chairman of Policy Research Council, People's New Party Secretary General, New Renaissance Party Deputy Secretary General, Liberal Democratic Party Vice- Chairman of Policy Research Council, New Komeito Party Vice Chair of Policy Research Committee, Democratic Party of Vice- Chairman of Policy Research Council, Your Party Chair for Diet Affairs Committee, Communist Party Chairman of Policy Research Council, Social Democratic Party Tokyo Electric Power Company (TEPCO) Prefectural Miyagi Ground Self-Defense Force US Army Prefectural Fukushima Ministry of Health, Labour and Welfare Prime Minister, New Zealand Prefectural Iwate Minister of Foreign Affairs, China Minister of Foreign Affairs and Trade, South Korea Prime Minister's Official Residence, Chair for Diet Affairs Committee, Democratic Party of Liberal Democratic Party Nissan Sendai Airport Toyota Trucking Association Ministry of Land, Infrastructure, Transport and Tourism Ministry of Internal Affairs and Communications Risk Management Center, Cabinet Office, President, House of Representatives of Sumitomo Metal Industries ese Consumers' Cooperative Union Red Cross Hiroshima University Ministry of Foreign Affairs and Trade, South Korea 9 Various types of centrality measure highlight importance of an organization in a network from different perspectives: degree centrality shows the extent to which an organization is connected with other organizations; betweenness centrality shows how important an organization is in terms of connecting other organizations; closeness centrality shows how easily an organization can reach other organizations in a network; and Eigenvector centrality shows how important, central, or influential an organization s neighbors are. 19

20 4-2. Three Distinct Networks Because the Great East Earthquake required responses to three different types of disasters earthquake, tsunami, nuclear it is possible to observe three distinct networks according to each policy area. In order to understand the response network from this perspective, each action identified in the newspaper articles was coded as involved in earthquake, tsunami, nuclear, or two or more of the policy areas. In this section, we review these three networks separately. Visualization maps of the three networks are found in Appendix. Table 6: Organizations in Three Distinct Response Systems by Funding Source Earthquake N(%) Tsunami N(%) Nuclear N(%) Public 519 (52.6%) 521 (53.7%) 218 (58.6%) Private 363 (36.8%) 344 (35.4%) 137 (36.8%) Nonprofit 104 (10.6%) 106 (10.9%) 17 (4.6%) Total 986 (100%) 971(100%) 372(100%) Table 7: Organizations in Three Distinct Response Systems by Jurisdiction Earthquake N(%) Tsunami N(%) Nuclear N(%) International 139 (14.1%) 143 (14.8%) 66 (17.7%) Regional 30 (3.0%) 26 (2.7%) 13 (3.5%) National 467 (47.4%) 468 (48.2%) 217 (58.3%) Prefectural 127 (12.9%) 112 (11.6%) 40 (10.8%) Municipal 73 (7.4%) 67 (6.9%) 14 (3.8%) Local 150 (15.2%) 155 (15.9%) 22 (5.9%) Total 986 (100%) 971 (100%) 372 (100%) Response Network for Earthquake There were 986 organizations responding to the situations triggered by earthquakes, and the density of the network turned out to be Over half were public organizations (52.6%), and almost half were national organizations (47.4%). Comparing the key entities in the overall network (Figure 5) and those in the earthquake network (Figure 6) the prefectural government of 20

21 Miyagi appears as one of the top ten key entities in the earthquake network, while it did not appear in the overall network. Figure 6: Key Entities in Earthquake Response Network Legend: dpj_sg: Office of Secretary General, Democratic Party of jp_pm: Prime Minister, japan_gov: ldp: Liberal Democratic Party pnp_prc: Chairman of Policy Research Council, People's New Party ldp_dsg: Deputy Secretary General, Liberal Democratic Party miyagi: Prefectural Miyagi cab_vmin: Vice Minister, Cabinet Office, ldp_pre: President, Liberal Democratic Party dpj_diet: Chair for Diet Affairs Committee, Democratic Party of Table 8: Top 10 Organizations in Four Centrality Measures (Earthquake) Degree Centrality Betweenness Centrality Closeness Centrality Eigenvector Centrality Chair for Diet Prime Minister, Minister of Foreign Affairs Affairs, Committee, Democratic Party of Office of Secretary General, Democratic Party of Chairman of Policy Research Council, People's New Party Deputy Secretary General, Liberal Democratic Party Prefectural Miyagi Minister of Foreign Affairs, China Minister of Foreign Affairs and Trade, South Korea Liberal Democratic Party Nissan Vice Tokyo Electric Risk Management Toyota 21

22 Chairman of Policy Research Council, New Komeito Party Secretary General, New Renaissance Party Vice Chair of Policy Research Committee, Democratic Party of Vice- Chairman of Policy Research Council, Your Party Chair for Diet Affairs Committee, Communist Party Chairman of Policy Research Council, Social Democratic Party Representative for House of Councilors, The Sunrise Party of Power Company Ground Self-Defense Force US Army Prime Minister, New Zealand Mayor, Yamada Town, Iwate Prefecture Vice Minister, Cabinet Office, Self-Defense Force () Center, Cabinet Office, Prime Minister's Official Residence, Trucking Association Ministry of Land, Infrastructure, Transport and Tourism Ministry of Internal Affairs and Communications Minister of Health, Labour and Welfare Defense Minister, Sumitomo Metal Industries Red Cross Ministry of Foreign Affairs and Trade, South Korea Consulate- General in Sendai, South Korea American Red Cross Showa University Response Network for Tsunami A network that emerged in response to tsunami involved 971 organizations, slightly less than the earthquake network. The density of was nearly the same as the earthquake network. The top ten key entities as well as composition by funding source and jurisdiction were also similar, with slightly more publ ic organizations, international, national, and local organizations in the tsunami network. There were 29 organizations that appeared only in the tsunami network, not in the earthquake network. These include municipal governments that suffered much damage by tsunami (Matsushima City, Shiogama City, Natori City, and disaster headquarters of Minami- Sanriku Town) as well as Active Fault and Earthquake Research Center, and Niyac Corporation, a private firm on energy procurement. 22

23 The similarities observed between the earthquake and tsunami networks can be understood as reflecting the fact that tsunami was considered to be part of countermeasures for earthquake in the disaster management plans in. Also contributing to this similarity is the fact that initial responses during the first three weeks were almost the same for these two types of disasters. These include provision of food, water, shelter as well as transportation. Figure 7: Key Entities in Tsunami Response Network Legend: dpj_sg: Office of Secretary General, Democratic Party of jp_pm: Prime Minister, japan_gov: ldp: Liberal Democratic Party pnp_prc: Chairman of Policy Research Council, People's New Party ldp_dsg: Deputy Secretary General, Liberal Democratic Party cab_vmin: Vice Minister, Cabinet Office, ldp_pre: President, Liberal Democratic Party dpj_diet: Chair for Diet Affairs Committee, Democratic Party of ldp_diet: Chairman, Diet Affairs Committee, Liberal Democratic Party Table 9: Top 10 Organizations in Four Centrality Measures (Tsunami) Degree Centrality Betweenness Centrality Closeness Centrality Eigenvector Centrality Office of Secretary Chair for Diet General, Democratic Party Prime Minister of Affairs Minister, Foreign Affairs, Committee, of Democratic Party of Chairman of Policy Research Council, Minister of Foreign Affairs, Liberal Democratic Party 23

24 People's New Party Deputy Secretary General, Liberal Democratic Party Vice- Chairman of Policy Research Council, New Komeito Party Secretary General, New Renaissance Party Vice Chair of Policy Research Committee, Democratic Party of Vice- Chairman of Policy Research Council, Your Party Chair for Diet Affairs Committee, Communist Party Chairman of Policy Research Council, Social Democratic Party Representative for House of Councilors, The Sunrise Party of Tokyo Electric Power Company Prefectural Iwate Ground Self- Defense Force Mayor, Yamada Town, Iwate Prefecture Vice Minister, Cabinet Office, Tokyo Metropolitan Police Department China Minister of Foreign Affairs and Trade, South Korea Risk Management Center, Cabinet Office, Prime Minister's Official Residence, Nissan Toyota Sendai Airport US Army Prefectural Iwate Trucking Association Ministry of Land, Infrastructure, Transport and Tourism () Ministry of Internal Affairs and Communications Minister of Health, Labour and Welfare Red Cross Ministry of Foreign Affairs and Trade, South Korea Consulate- General in Sendai, South Korea American Red Cross Showa University ese Association of Dialysis Physicians Response Network for Nuclear Threat The response network for nuclear threat shows a distinct feature compared to earthquake and tsunami networks, as would be expected from the way nuclear disasters were treated in the disaster management law and plans. It is a smaller network of only 372 organizations, with slightly higher density than the others (8) and smaller average distance. Reviewing the visualization diagram of the network (Appendix), we see several isolated groups of organizations interacting/communicating with each other, with a cluster of organizations around the Prime Minister of and the ese government. A smaller number of nonprofits were involved 24

25 in the nuclear network, and nearly 60 percent of the responding organizations were national organizations. Nonetheless, key entities included Tokyo Electric Power Company (TEPCO), the private firm operating the Fukushima Daiichi Nuclear Power Plant, as well as the prefectural government of Fukushima where the plant is located. While the National Safety Commission (NSC) appears important in terms of betweenness centrality and closeness centrality, the Minister of State for Disaster Management does not appear at all. Figure 8: Key Entities in Nuclear Response Network Legend: jp_pm: Prime Minister, tepco: Tokyo Electric Power Company (TEPCO) dpj_sg: Office of Secretary General, Democratic Party of nrp_sg: Secretary General, New Renaissance Party pnp_prc: Chairman of Policy Research Council, People's New Party japan_gov: fukushima: Prefectural Fukushima mhlw: Ministry of Health, Labour and Welfare cab_sc: Chief Cabinet Secretary, sdf_gr: Ground Self-Defense Force Table 10: Top 10 Organizations in Four Centrality Measures (Nuclear) Degree Centrality Betweenness Centrality Closeness Centrality Eigenvector Centrality Office of Secretary General, Democratic Party of Chairman of Policy Research Council, People's New Party Tokyo Electric Power Company Prime Minister, NTT East (telephone) Ministry of Land, Infrastructure, Transport and Prime Minister, Hiroshima University 25

26 Secretary General, New Renaissance Party Deputy Secretary General, Liberal Democratic Party Vice- Chairman of Policy Research Council, New Komeito Party Vice Chair of Policy Research Committee, Democratic Party of Vice- Chairman of Policy Research Council, Your Party Chair for Diet Affairs Committee, Communist Party Chairman of Policy Research Council, Social Democratic Party Representative for House of Councilors, The Sunrise Party of Ground Self-Defense Force Minister of Economy, Trade, and Industry Governor, Fukushima Prefecture Prefectural Fukushima Ministry of Health, Labour and Welfare Defense Minister, Chair of Nuclear Safety Commission Tourism Governor, Tokyo Prefecture Nuclear Safety Commission President, Tokyo Electric Power Company European Commission Tokyo Metropolitan Police Department Agency for Natural Resources and Energy Bank of Tokyo- Mitsubishi UFJ Mitsui Sumitomo Bank President, United States Secretary of Energy, US President, House of Representatives of President, House of Councilors of National Institute of Radiological Sciences Chair for Diet Affairs Committee, Democratic Party of Office of Aizawa, Member of Liberal Democratic Party Governor, Bank of Reviewing the three networks separately, we observe that networks that emerged in response to different policy areas involved in the Great East Earthquake were quite different. As implied in the Disaster Countermeasures Basic Act and the Basic Disaster Management Plan, the networks for earthquake and tsunami response shared similar properties, but included slightly different sets of organizations. The network responding to the nuclear threats revealed a completely different profile, both in terms of the structure of the network as well as composition of participating organizations. Only 283 organizations were identified as overlapping in all three networks, of which 57.6% were public and 62.9% were national. 26

27 4-3. Increasing Communication Efficiency in Fragmented Networks The three policy areas involved in responding to the Great East Earthquake earthquake, tsunami, and nuclear disaster require different expertise. However, with the distinct events evolving virtually simultaneously while affecting each other, the three policy groups ideally would need to work together for effective and efficient response to the disaster. Given the way the Disaster Countermeasures Basic Act and the Basic Disaster Management Plan are elaborated, we would not expect such communications across the three policy groups. However, by examining the interactions among organizations in the emerging network, we discover that the responding organizations voluntarily interacted across the policy areas in coping with the actual event, indicating an emerging small world network. One method to measure efficiency in communication is to calculate the ratio of clustering to distance among organizations (Comfort, Siciliano, and Okada, 2010). A small world network (Watts and Strogatz, 1998; Newman et al. 2006) is a subset of a larger network that demonstrates a short distance among other organizations in the network, coupled with a high clustering coefficient in comparison to randomly distributed organizations. The emergence of a small world network indicates the existence of a subset of organizations that transmit information or materials more quickly and efficiently within the system than if the same information or resources were distributed randomly throughout the entire network. The measures for clustering and distance among the nodes in the network are cited in comparison to similar statistics generated for random graphs. Previous studies have identified 4.75 as the threshold for a network having the small world property (Kilduff et al. 2008). Calculating the small world ratio for the three distinct networks in ORA, we do not see the evidence of small world properties. This implies that efficiency in communication within each policy group was not as high. However, in the overall network of all 1,101 organizations, the ratio rises to , almost reaching the threshold of This gap implies that bringing together the three policy groups increases efficiency in communication, reflecting the reciprocal interdependent nature of the disaster that struck in March 11, What is impressive is that these cross-policy area communications were self-organized, with little guidance from the Act or the Plan. There were indeed organizations that overlapped in more than one policy area; the small world ratio obtained here implies that these overlapping organizations contributed to 27

28 bridging the three policy groups together. Had the Disaster Countermeasures Basic Act and the Disaster Management Basic Plan assumed the interdependence demanded in responding to the three disasters that occurred virtually simultaneously thus encouraging organizations in different policy areas to communicate the small world ratio might have been even higher, enhancing efficiency in communication in disaster management. Table 11: Small-World Ratio 10 Earthquake (ORA) Earthquake (UCINET) Tsunami (ORA) Tsunami (UCINET) Nuclear (ORA) Nuclear (UCINET) Overall (ORA) Overall (UCINET) Node Count Density Number of Ties Clustering Coefficient Average Distance Average Clustering (Random Graph) Average Distance (Random Graph) Clustering Ratio Distance Ratio Small World Ratio Conclusion: Theory vs. Practice in Interpreting the March 11, 2011 Disaster What struck on March 11, an earthquake of magnitude 9.0, followed by a tsunami that triggered a nuclear threat is a perfect test case for policy makers to ponder the challenge of planning for rare and extreme events. Because Black Swan events are incidents that go beyond one s imagination, no one can determine when preparations for such events are adequate. The 10 In addition to small world ratios calculated using ORA, this chart also demonstrates ratios calculated with UCINET. ORA and UCINET uses different versions of the formula to calculate clustering coefficient. While UCINET does not include values for nodes that have only 1 tie or no ties in calculating clustering coefficient at the network level, ORA assigns 0 as clustering coefficients for these nodes. As such, clustering coefficients produced in ORA are always smaller than those produced in UCINET. 28

29 question to be asked, therefore, was: to what extent and how should investments be made in preparing for rare events? Review of the Disaster Countermeasures Basic Act as well as the Basic Disaster Management Plan revealed that was well prepared for earthquakes, the type of disaster that the country had frequently experienced in the recent years. However, had not given sufficient attention to tsunami, merely treating it as part of the areas in need of countermeasures in case of an earthquake. Nuclear disasters were treated as a completely separate type of disaster in planning and response. Networks of organizations that emerged in response to the disasters on March 11, 2011 reflected the characteristics of the Act and the Plan. Networks of earthquake and tsunami appeared to have similar structure and composition of organizations, while the response network for the nuclear threat revealed a strikingly different structure and composition of organizations. However, differences were observed between plans and practice. While the plans had assumed public organizations to be the main actor in response, a number of private organizations and nonprofits formed response networks. Similarly, while organizations in the plans were mostly national with some prefectural and municipal in its jurisdictional attribute, international and local organizations came to play a role in response networks. The networks responding to each of the three disasters (especially between the earthquake/tsunami and nuclear networks) were quite fragmented, given the differences in density, compositions of organizations and central organizations. Looking at the small world ratio, communications across the three policy areas appeared to have taken place, despite the lack of such guidance in the disaster management plans in. Such a self-organizing trend meant an increase in efficiency in communication and knowledge sharing, potentially leading to better response to the disasters. One lesson learnt from the case, thus is the need to recognize the possible occurrence of multiple disasters simultaneously, and to enhance and facilitate communications among different policy groups. While preparation can never be considered adequate, doing so contributes to improving response to rare and extreme events. The continuing question is what have we learned from this catastrophic event and how can we improve disaster management to cope with the increasing likelihood of potential interaction of natural hazards with large-scale sociotechnical systems that are designed to support 29

30 and sustain human communities. Perrow s argument regarding the interdependence of largescale technical systems and their inherent vulnerability to failure cannot be denied in long-term policy planning. The next question for policy makers is whether other sociotechnical mechanisms that support collective learning and action can be adapted to counter the always present risk of failure. The use of a well-designed information infrastructure to support collective learning offers promise, as well as increased recognition that organizations could constructively participate in collaborative rationality to reduce risk, not simply respond after the incident has occurred. The results from the small world network analysis suggest a potentially productive approach. If such small worlds could be productively developed around known risks to improve practices of communication and coordination among organizations that would necessarily be involved in response, the likelihood of increased awareness, improved performance in practice, and more timely response would increase. As with every disaster experience, is beginning to revise the disaster management plans and systems to better prepare for what will happen in the future. The Central Disaster Management Council has established a special committee to analyze and incorporate the lessons learned from the Great East Earthquake in April. Unfortunately, most of the recommendations have been directed towards enhancement of countermeasures for tsunami, and revision of envisioned scale of earthquakes including the possibility of multiple predicted earthquakes occurring simultaneously (Central Disaster Management Council, 2011b). Further, to our knowledge, there has been little discussion on evaluating communication and information flow during the response for the Great East Earthquake. This paper initiates such discussion and proposes the need to enhance collaborative participation and communications among various policy groups with the use of information infrastructure, leading to a creation of knowledge commons. Doing so will not only increase efficiency in communication in disaster management and thus reduce risks, but also enhance the quality of response when rare and extreme events do take place. 30

31 References Cabinet Office of. (2011a). White Paper on Disaster Management (in ese). Available at Cabinet Office of, Director General for Disaster Management. (2011b). Disaster Management in (in ese). Available at Central Disaster Management Council. (2011a). Central Disaster Management Council Member List. Available at Central Disaster Management Council. (2011b). Report from the Specialized Committee on Earthquake and Tsunami Countermeasures Based on Lessons Learnt from Earthquakes off the Coast of Pacific Ocean in Tohoku Region (in ese). Reports available at Comfort, L. K., M. D. Siciliano et A. Okada (2010). Risque, Résilience et Reconstruction: Le Tremblement de Terre Haïtien du 12 Janvier Télescope, Vol.16(2), pp Hess, C. and Ostrom, E. (2006). Understanding Knowledge as a Commons: From Theory to Practice. Cambridge, MA: MIT Press. Innes, J.E. and Booher, D.E. (2010). Planning with Complexity: An Introduction to Collaborative Rationality for Public Policy. New York: Routledge. Atomic Energy Commission (AEC) (ed.). (2009). White Paper on Nuclear Power 2009 Edition (in ese). Tokyo: Energy Forum. Kanazawa, F., Obara, H., Oote, M., Yamamoto, T., Uesaka, K. (2007). Standard Specifications for a Disaster Information Platform (in ese). Technical Note of National Institute for Land Infrastructure Management. Available at Kilduff, M., Crossland, C., Tsai, W. and Krackhardt, D. (2008). Organizational Network 31

32 Perceptions Versus Reality: A Small World After All? Organizational Behavior and Human Decision Processes. Vol.107(1), pp Kusakabe, T., Sanada, A., Uekasa, K., Yamamoto, T., Kawase, K., Shimada, H. (2007). Notification and Sharing of Disaster Information by Disaster Information Sharing Platform (in ese). Technical Note of National Institute for Land Infrastructure Management. Available at Miyako City. (2011). Summary of Tsunami Countermeasures in Taro District (formerly Taro Town). Document obtained during field visit, November 1, National Policy Agency (NPA). (2011). Updated Information on Victims and Policy Responses to Earthquakes off the Coast of Pacific Ocean in Northeast (in ese). Available at Newman M., Barabasi A.L., Watts, D.J..(2006).The Structure and Dynamics of Networks. Princeton, NJ: Princeton University Press. Nobre, F.S., Tobias, A.M., and Walker, D.S. (2009). The Impact of Cognitive Machines on Complex Decisions and Organizational Change. AI and Society, Vol.24(4), pp Perrow, C. (2006). Disasters Ever More? Reducing U.S. Vulnerabilities. Pp in Handbook of Disaster Research, edited by Havidán Rodríguez, Enrico L. Quarantelli, and Russell R. Dynes. New York: Springer. Seibido Mook: Reading the Great East Earthquake on Maps. (2011). Tokyo: Seibido Shuppan. Taleb, N.N. (2007). The Black Swan: The Impact of the Highly Improbable. New York: Random House. TEPCO: Evidence on Assuming Tsunami over 15m Claim for Unexpectedness Declined (in ese). (2011, August, 25). Yomiuri Newspaper. The Great Earthquake Regretting the Inability to Envision Seismology Society Declares a New Start (in ese). (2011, October 15). The Nikkei Newspaper. 32

33 Watts D.J. and Strogatz, S.H. (1998). Collective Dynamics of 'Small-World' Networks. Nature, Vol.393 (6684), pp

34 Appendix: Network Maps (Isolates Hidden) Legend Jurisdiction Red: International Blue: Regional Green: National Light Blue: Prefectural Yellow: Municipal Orange: Local Funding Source Circle: Public Triangle: Private Square: Nonprofit 34