Improving communication resilience for effective disaster relief operations

Abstract

The objective of this paper is to identify strategies to improve the resilience of interagency communication between relief organizations and the community when dealing with an emergency. This research draws from frameworks including information theory, organization design, and how the private sector has learned and evolved from the challenges of information flow to provide guidance to disaster relief agencies. During times of emergency, private organizations as well as public authorities must coordinate in real time to create an effective response. When coordination is absent, failure results, as was seen after Hurricane Katrina and the Haiti Earthquake. Using data that the authors collected immediately after these disasters, two case studies of systemic failure are presented to extract lessons that might be used to improve communication resilience through coordination between parties in humanitarian relief operations. Recent emergency response trends are identified, and the paper argues that the persistence of response failures is not surprising, in part because response organizations normally operate independently, and their operations evolve at different rates. As a result, the organizational interfaces that enable rapid integration during a disaster naturally degrade and may be weak or absent. Integrating the literature on information processing theory and organization design with the data from the two case studies, the paper proposes that increasing the resilience of disaster response systems can be achieved by (1) improving the interoperability and information flow across organizational boundaries; (2) increasing the synergies between organizations on adapting new technology such as social media for the coordination of structured and unstructured data for use in decision-making, and (3) increasing the flexibility of relief organizations to use external resources from areas not affected by disasters on an opportunistic basis. The paper concludes by discussing resilience enhancing solutions including boundary spanning investments and argues that effective emergency response does not result from sporadic or intermittent efforts but rather requires sustained investment, continuous monitoring, and data collection.

1 Introduction

Over the last decade, the frequency of man-made and natural disasters has increased (Whybark et al. 2010), creating unprecedented demand for humanitarian assistance (Van Wassenhove and Pedraza Martinez 2012). A major factor that increases the amount of property damage and the number of human lives lost is that the communication systems in place to respond to disaster have not evolved as rapidly as the underlying technologies. This is unfortunate because numerous technological advancements, such as wireless cellular communications systems, have become globally available and provide significant increases in capability. In this paper, we argue that a lack of communication and coordination prevents organizations from responding effectively and from focusing on the right issues in a disaster. For example, Scholtens (2008) shows that when the necessary centrally controlled collaboration in a disaster is absent, relief operations will be ineffective. However, such collaborations are predicated on voluntary actions and are supported by investments. Thus, the thesis of this paper is that the failure to coordinate and incorporate information flows from different sources is critical and underlines the weakness of relief operations in the aftermath of a disaster. This failure can be mitigated by improving communication resilience.¹

The National Academy of Sciences defines disaster resilience as “the ability to plan and prepare for, absorb, recover from, and adapt to adverse events” (Larkin et al. 2015; Linkov and Palma-Oliveira 2017). This paper adopts a definition more closely aligned to the topic of communication resilience in the context of a disaster: “resilience offers the capability to better review how systems may continually adjust to changing information, relationships, goals, threats, and other factors to adapt in the face of change—particularly those potential changes that could yield negative outcomes” (Linkov et al. 2016). Thus, this paper is bounded by, and consistent with, the three phases of resilience management defined in the European Resilience Management Guidance (ERMG) developed by the RESILIENS team. The phases are as follows: (i) prior to a disaster, agencies must prepare, prevent and possibly protect; (ii) during a disaster, response agents must mitigate, absorb and adapt; (iii) and following a disaster, respond, recover and learn (Antunes et al. 2017). Reliability and resilience strategies are now critical in systems such as energy (Deluque et al. 2018; Yuan et al. 2016; SEPA 2018; Shittu et al. 2015; Shittu and Baker 2009), transportation (Litman 2006), and health (O’Sullivan et al. 2013) with resilience capacities captured across three dimensions including adaptive, absorptive, and restorative (Francis and Bekera 2014).

This paper draws lessons from the literature on information theory and organization design and compares the communication between different government and private groups focusing on the use, or lack, of wireless and cellular communication systems after Hurricane Katrina and the Haiti Earthquake. The authors collected, and analyzed, primary and secondary data in the aftermath of these disasters that show how the communication failures occurred and make recommendations for how to improve communication in future disasters. In the context of this paper, effective communication occurs when what is stated by the sender is exactly what is received and understood by the receiver (McLuhan 2008; McLuhan and McLuhan 2011). Consequently, the construct of “lack of communication” is important since it refers to when the receiver does not get the message sent by the source. Thus, whether communication occurs may be impacted by factors such as information overload, information quality, information links, biases (both implicit and explicit), and culture.

To provide a concrete example, consider the tsunami that struck in Ukedo, Japan in 2011. Of the village’s 1800 residents, 10% were killed due to the unnecessary evacuation that was ordered by the authorities. Some might have been saved, but authorities’ lack of coordination resulted in some questionable decisions. When the first of three reactor buildings at the nearby Fukushima Dai-ichi nuclear power plant blew up, the authorities’ focus shifted to evacuating the living rather than searching for survivors and survivors were ordered to drive to evacuation shelters further away. Unfortunately, this put them directly in the path of the nuclear plant’s plume of radioactivity and deprived the elderly of the care that was needed. This and similar decisions indicate a lack of coordination between different organizations within the Japanese response system (The Economist 2012). This failure to accurately communicate known danger parallels errors that Goldsteen and Schorr (1991) identified immediately after the 1979 Three Mile Island nuclear plant emergency—these were communication errors that led to lack of knowledge about the relative risks. These errors amplified the consequent losses reported in these disasters. Thus, people lost trust and concluded that they were either being lied to or that the responsible organization was incompetent.

The 2010 Haiti Earthquake killed more than 200,000 people, injured many more, and left more than a million people without shelter. In scale, it was the worst disaster Haiti has experienced. The relief and recovery efforts instituted were the largest since the Indian Ocean tsunami of December 2004 (IOB 2011). Unfortunately, the response to the disaster was disjointed and lacked coordination across international relief agencies and local responders (VanRooyen 2013). Ironically, researchers had observed the lack of effective response for many years. The relatively long interval between disasters in any one area prevents the joint evolution of independent response systems, because the investment costs are high and system usage is low. An appealing solution to the lack of disaster response coordination might be the development of an integrated one-size-fits-all disaster response system. However, too much integration creates an unwieldy interdependent system that is slow to evolve and slow to integrate new technologies such as social media networks (Waibel et al. 2009). This paper uses Hurricane Katrina and the Haiti Earthquake as case studies to understand the application of boundary spanning theory in disaster response coordination. Boundaries define and distinguish an organization from others, and boundary spanning is the link between the environment in which an organization operates and the organization (Aldrich and Herker 1977).

This research makes a major contribution—it offers strategies for enhanced future deployment and integration of emerging resilient communication channels into existing critical infrastructures. This research unpacks the value proposition in dynamic learning and interoperability between independent organizations along the path of adaptive and resilient systems strengthened by information fidelity, feedback, and investment in critical data for recovery. The approach offers the potential to increase the performance of future response and relief operations particularly on interoperability, swift trust, and the management of the challenges created by emergent providers on coordinating and managing information flows. The paper recognizes the need for greater visibility of such flows and how the challenges of information flow can be addressed in an environment where there is a need for greater transparency.

The rest of this paper proceeds as follows: Sect. 2 provides a succinct but comprehensive review of related literature. Section 3 provides a theoretical framework that underlies the analysis of the case studies presented in Sect. 4. The analysis of the cases in Sect. 4 also highlights the data, sources of data, and the methodology employed. Section 5 discusses the way forward based on the highlights from the case study, and Sect. 6 concludes.

2 Literature review

This review draws from four different research streams. First, we highlight studies that identify how adaptive systems require integration based on the extant literature on Hurricane Katrina and the Haiti Earthquake. Second, we examine the literature on organizational design for emergency operations and identify the missing gaps. Third, we explore how communication infrastructures and interoperability might adapt to the emerging use of social media for disaster mitigation. Lastly, we highlight the relevance of knowledge management systems in the development of effective disaster and relief operations.

2.1 Complex adaptive systems

This section provides a review on how coordination in communications leads to the need adaptive systems for emergency response. Particularly, the relevance of systems theory to integrate interdependent and autonomous entities is discussed. The U.S. President’s Commission on Critical Infrastructure Protection (PCCIP 1997) lists telecommunications, electric power systems, and emergency services as three of the eight most critical infrastructures for maintaining the nation’s defense and economic security. These infrastructures have complex interactions that change over time, making them complex adaptive systems. Thus, disasters create a cascade of disruptions among these interdependent infrastructures (Comfort 1999), such that failure in one system triggers failure in the other (Santos et al. 2014; Comfort et al. 2004a). This problem is exacerbated by the number and diversity of actors in the coordination of humanitarian relief operations (Balcik et al. 2010), because the challenges of inter-organizational communication impede coordination between these actors (Comfort et al. 2004b).

James and Cubano (2012) discuss the collaboration between the Department of Defense (DOD) and a non-governmental organization (NGO) during the Haiti earthquake crisis by highlighting the factors that worked and the hurdles that the partnership experienced and suggests combined training and exercises that consider the differences of the organizations, and cooperation for organizational synergy. Ferris and Ferro-Ribeiro (2012) detail the difficulties of collaboration between camps of international responders and the spontaneous settlements in urban areas in the aftermath of the Haiti earthquake. Nolte and Boenigk (2011) reinforce the importance of the partnership between the public and non-profit organizations in times of disaster response. With the aid of a process model, the paper evaluates intersectoral partnerships derived from theory and offers a systematic literature review of emergency, public, non-profit, and network research. The paper finds that communication, trust, and experience are the most important partnership inputs. Also using public–private partnerships, Chen et al. (2013) highlight how the fragmented ownership of disaster response leaves a void for response responsibility. The paper advocates for increased informal control and new coordination structures to facilitate post-disaster recovery and learning. Comfort et al. (2011) explore the tensions among resilience, entropy, and efficiency in interactions between organizations operating in response to the Haiti Earthquake within the first three weeks of the disaster. The paper finds the three factors in interpreting the patterns of performance for constructive action include knowledge of seismic risk, public investment in network and information technologies and the capacity of the community to learn. The paper recommends that timely intervention in information flow can be used to minimize entropy, increase efficiency, strengthen resilience and ensure a smooth transition from response to recovery (Comfort et al. 2011).

Systems theory provides a useful perspective for emergency managers to view the relationships between interdependent, yet autonomous, learning entities, and becomes a tool to prepare for and respond to disasters (Pine 2010; Bharosa and Janssen 2009). Altay and Labonate (2014) analyze the fragmented nature of response and the use of hierarchical models of information management during the Haiti earthquake and concludes that permeability, information ownership, and information integrators are critical requirements for effective response operations. Altay and Pal (2013) evaluate the information processing aspect of the UN cluster approach to disaster management and suggests that if cluster leads act as information hubs, information reaches intended targets faster, enabling a prompt humanitarian response. It also outlines coordination and the effect of cluster leads acting as information filters to prevent chaos and proper dissemination of information for effective resource usage.

Day (2014) argues that these systems are complex adaptive supply networks that facilitate community resilience. Thus, an approach to improve disaster response is to deploy effective supply chain and logistics operations (Day et al. 2012; Tomasini and Van Wassenhove 2009; Van Wassenhove 2005). The complexity of these operations and systems is due to their interdependence with autonomous and potentially unpredictable actors/responders; however, it is not only the ability to adapt that is important, but also the speed with which entities respond to changes around them. Bolton and Stolcis (2008) find agents’ behavior effective in some ways yet ineffective in others. For example, government responses to September 11, 2001 attacks and Hurricane Katrina showed some innovative and proactive decision-making. Nevertheless, they were cases of poor planning, poor execution, administrative rigidity, and faulty application of good emergency management practices, because the collaboration and horizontal communication called for in fast-breaking crises were missing (Cutter and Gall 2007).

We fill a gap in the literature by identifying effective emergency preparedness and management prerequisites including continuous exercises that are self-organizing and self-adaptive as the agents, organizations, and individuals respond to external stimuli. We recognize that learning across multiple entities depends on how tightly coupled the systems and their feedback loops are, their incentives, and their response behaviors.

2.2 Organizational design for emergency operations

The review in this section highlights the role of social partnerships across organizations for effective disaster response. The influence of role structure on coordination is highlighted using recent studies on the Haiti Earthquake and Hurricane Katrina. Pyles et al. (2017) examine if disaster impacts and social capital (social trust and civic engagement) are associated with disaster recovery participation and compares predictors of such engagement in Haiti earthquake and Hurricane Katrina. The paper finds that the advantages of local and community participation include effective identification of needs and issues, improvement of community resilience, and greater efficiency in the use of resources. Chen et al. (2013) highlight the importance of partnerships between various organizations in the public, private, and non-profit sectors to improve disaster resilience. The paper draws insights from common operational and administrative issues associated with such partnerships during disaster management and illustrates how the severity of these problems (as a function of the uncertainty and complexity of the tasks) recommends adaptive coordination and control measure.

In organizational terms, coordination is the management of dependencies between activities (Crowston 1997) that requires long term establishment for success. In the context of relief operations, coordination refers to the relationships and interactions between different actors operating within the relief environment (Balcik et al. 2010). As coordination need increases, the uncertainty facing organizations increases (Galbraith 1973). The four parameters that define temporary organizations are time, task, team, and transition Lundin and Sderholm (1995); however, if we adopt Bechky’s view (2006), we might decide that the portrayal of the operations of relief organizations as ephemeral and unstable is incomplete. Instead, we believe Bechky would suggest that effective coordination takes place when temporary institutions are enduring entities that combine in novel ways for each new disaster. By combine, we mean how role structure and negotiation are the drivers that underline the order of coordination between temporary organizations. We exemplify novel ways to interpret the roles that are enacted by basing such roles on the social arrangement between the different ephemeral organizations, and the constructive ways they come together. Coordination is also hampered by competition for limited resources amongst relief agencies. However, in the time between Katrina and Haiti, the UN introduced nine clusters for coordination around specific themes. Within each cluster, there is a dedicated agency that functions as the “provider of last resort.” For example, the Emergency Telecommunications cluster that includes the UN Office for the Coordination for Humanitarian Affairs (OCHA), the United Nations Children’s Fund (UNICEF), and the World Food Program (WFP) aided the coordination improvement observed in the Haiti response.

Parker et al. (2009) focus on the inadequacy of coordination amongst federal, state, and local authorities that led to the lack of preparedness, inadequate collective response to a long-predicted, upper-category Hurricane Katrina. The paper argues for sustained investment to improve community resilience. Olejarski and Garnett (2010) discuss the behaviors that significantly contribute to conflict and therefore communication failures using Hurricane Katrina and suggests the need for elected officials and top management to understand crisis management archetypes and to conduct civic engagement and outreach initiatives. Boin and McConnell (2007) explore how public authorities can effectively prepare to cope with breakdowns of critical infrastructure systems by preparing first responders to work with communities in joint preparation and training to facilitate coordination and effective communication. Curtis (2016) uses social network analysis to examine how organizations in the response to Hurricane Katrina were integrated into the relief effort following the disaster and suggests that organizations with valuable resources must be included in the networks of responders (Curtis 2016). Earlier, Curtis (2015) explores how coordination and communication between the government and service organizations led to the poor outcomes for the communities impacted by Hurricane Katrina and suggests that increasing inter-organizational communication, preparedness, timely meetings between organizations, and establishing clearly defined roles for organizations improve disaster response.

This paper provides the missing link by arguing that how ephemeral organizations interact with traditional relief agencies is critical to the success of relief operations. That success, however, is dependent on the effectiveness of the underlying information platforms that allow the relaxation of organizational boundaries as discussed in the next section.

2.3 Information technology and boundary spanning

This section highlights some of the literature on the prerequisite for an information exchange platform required to enhance the adaptive capacities of organizations to coordinate for effective response. A goal of organizational design for managing complex tasks is to enable organizations to overcome barriers to the transfer of tacit knowledge across cultural and structural barriers so that sufficient information is available to arrive at good decisions (Huber 1991). Scott (2008) indicates that the mechanisms for managing information transfer within firms can be divided into two approaches: basic (rules and programs, schedules, departmentalization, hierarchies, delegation, and use of “task objects”); and advanced (product or matrix organizations, slack resources, use of information technology). These approaches are supplemented by a variety of lateral relations such as those described by Lawrence and Lorsch (1967). Similarly, the literature has identified several mechanisms for managing information transfer between firms including increasing the absorptive capacity of one or both firms (Cohen and Levinthal 1990), employing boundary spanners, creating strategic alliances, or participating in institutionalized practices and procedures (Thompson 1967; Scott 2008). Although most disaster response organizations are non-profit or governmental, they face most of the same coordination issues as the private firms detailed in the literature. Thus, we close a gap in the literature by harnessing firm-level organizational coordination and integration strategies in framing a theory for relief operations. Rivera and Nickels (2014) discuss the inadequacy of government and local structures put in place to mitigate disaster risks and the rise of reliance on other types of organizations. For example, the paper examines how a church-led organization raised social capital in the event of decreased civic trust and suggests that an empirical analysis between faith-based organizations and other formal disaster response and recovery agencies be performed to allow social capital and legally responsible organizations perform such activities more efficiently.

Yates and Paquette (2011) present a case study of the Haiti Earthquake that shows how social network-enabled knowledge management systems helped to overcome syntactic, semantic, and pragmatic boundaries. For example, bridging a syntactic boundary implies that two or more parties have a common language for communication. Overcoming the semantic boundary requires the absence of syntactic boundary so that ambiguity in the interpretations of the given data is limited or absent. Lachner and Hellwagner (2008) suggest solutions using new concepts and techniques from Service Oriented Architectures, Semantic Systems, and Artificial Intelligence. Standardization efforts have led to Common Alerting Protocol (CAP) and the Emergency Data Exchange Language (EDXL). These need to be extended for coordination on the strategic and tactical levels to become more efficient and effective (Iannella and Henricksen 2007).

This paper argues that the strength of information processing through activities such as geocoding and cross-communication underscores the fidelity of information required to achieve communication resilience. The identified organizational boundaries, i.e., syntactic, semantic, and pragmatic, are better bridged when the coordinating infrastructure is grounded in the emerging IT platforms. Specifically, this paper suggests that internet and mobile computing tools have created the medium for all actors in response operations.

2.4 Knowledge management systems

This review closes with exploring some of the papers that highlight the relevance of knowledge management as a critical element of sociotechnical systems theory. Disaster response operations depend on effective knowledge management systems. This is because disaster response involves a coordinated set of interactions between individuals and relief agencies. These agencies have roles, applicable expertise, and deliverables that vary by disaster, each of which has a distinct geographic and socio-economic environment. There are many common needs across disaster responses that include information sharing, transport, and medical/social professionals (Yates and Paquette 2011). Knowledge sharing has become a key aspect of sociotechnical systems (Broniatowski 2018) such that quick and efficient information retrieval and sense-making support adaptation and improvisational survival strategies (Wachtendorf 2004). However, for effective information retrieval, data collection and storage must align with universally defined standards (Mock and Garfield 2007) or rapid/real-time learning strategies to identify these in the face of crisis. Liu and Kim (2011) expose the subtleties of information consumption patterns during disaster. They suggest that organizations rely on traditional media versus social media. When the crisis is framed as a health disaster, reliance on traditional media is higher. However, when the problem is framed as a general crisis, both means are equally employed. Regardless of these differences, Ediger et al. (2010) show that the social interaction graph induced by public Twitter messages demonstrates Twitter as an effective news disseminator during disasters.

This paper closes the gap on communication resilience enhancement by suggesting that knowledge management systems require investments to limit incidences of false positives or negatives and improve the integration between the different data sources in platforms such as Twitter and Facebook.

3 Theoretical framework

Communication resilience in a disaster may be viewed as dependent on a tripod stand. The three defining features include information fidelity, feedback, and sources and type of data and investment. Figure 1 shows the three legs of the tripod representing the theoretical framework in this study. First, it is imperative that information fidelity informs the response strategy. However, the evolution of disparate information and communication infrastructure across the boundaries that separate relief agencies creates gaps in interoperability. Fine (1999) introduced the idea of industry clock speed in a comparison of multiple industries, including computers, automobiles, and media. Critically, the information flows within firms and across firm boundaries must change as industry and firm structure change. For example, in the early 1900s, there were many automotive manufacturers, each of which specialized in certain parts of the industry, such as Fisher Body, which in 1916 supplied bodies to assemblers such as Abbot, Buick, Cadillac, and Chalmers.

Fig. 1

Theoretical framework to enhance communication resilience

With the rise of mass manufacturing, firms vertically integrated to gain greater control over their supply chains. At Ford’s River Rouge complex, raw materials such as iron ore went in at one end, and finished products came out at the other. Over time, this structure grew extremely complex and inefficiencies emerged. By the 1990s, North American and European firms began to follow their Japanese counterparts and started the process of outsourcing larger fractions of vehicle value to a global supplier base that includes firms such as Bosch and Denso (Womack et al. 1990). To make such outsourcing effective, the firms had to restructure to efficiently interact with such highly capable suppliers.

One of the challenges of disaster management becomes the problem of free information flow across organizational boundaries that ultimately affect information fidelity and availability. Two of the most common arrangements to increase inter-firm processing capacities through lateral relations are as follows: (1) dedicated boundary spanners who can help interpret and mediate between firms (Parker and Anderson 2002); and (2) common venues for personnel from different firms working together, also called co-location. Either of these options has improved coordination between organizations (Schneider and Bowen 2010), increased the absorptive capacity of firms (Cohen and Levinthal 1990), and spanned structural holes, thus brokering knowledge among firms (Uzzi 1996) through the transfer of explicit and tacit knowledge among organizations. Further, the transfer of practices facilitates joint problem solving (Uzzi 1997). In disaster contexts, geocoding may be the proxy for co-location. Information fidelity may be improved with geocoding. Geocoding provides visualizations for geographical coordinates corresponding to a location and the needs of that location.

Second, feedback during a time of crisis is critical to the outcome of relief operations by both relief organizations and ephemeral or ad hoc agents in a disaster. Carley and Harrald (1997) discuss how organizational learning in theory and in practice are different, leading to the development of plans that are not viable. They find that feedback, a necessary component for learning, is often not available to, or wanted by, disaster response organizations. For example, where objective performance feedback supports organizational learning, response organizations are more disposed to subjective performance feedback from the media, and this decreases their ability to learn. In this context, experience is gained before, during, and after the occurrence of a disaster. In the absence of feedback, problems ranging from inadequate crisis mapping to trust issues and boundary conflicts make relief operations difficult to evaluate. Thus, the feedback loop we propose will provide closure and support progress monitoring.

Third, social media has emerged as a platform for knowledge. Starbird and Palen (2011) empirically consider the emerging role of the “digital volunteer” and identify how microblogging platforms serve as a new arena for self-organizing in the context of disasters. Social media or other tools for backchanneling (Sutton et al. 2008), that are seen to have the disadvantage of crowdsourcing which risks blurring behavior, are also an important avenue for volunteerism. Thus, strengthening the link between agencies, national governments, and individuals in response coordination through social media deployment that transcends traditional communication methods is a prerequisite for effective crisis management. In developing dynamic models that engage emerging communication media, the pace of their evolution through investment-driven learning must be synchronized not only with established communication modules, but also with the scope of relief operations. Such learning formalisms have been modeled in energy systems (Shittu 2014). The social media component ties back to the mechanism for feedback as a key component for successful relief operations. A prerequisite for such success relies on a resilience tool like Twitter and other social media applications such as Facebook and Google+. Reducing false positives or negatives, or noise, in the feedback loop can be achieved by increasing the weight of tweets that have been retweeted (Verma et al. 2011). The conundrum lies in how to integrate the data provisions on these platforms particularly structured data sought and provided by traditional organizations, and unstructured data provided by ad hoc or ephemeral entities. In this context, structured data are defined as information that has a high degree of organization in a well-defined relational database that is readily searchable, whereas unstructured data are essentially the opposite—search is tedious, and compilation is time consuming.

Big data can be leveraged to set up rapid archiving and application of visualization analytics in real time. To date, these activities have not been systematically implemented or invested into. As discussed above, a natural partnership between the traditional disaster response community and academia might be considered. In addition to a systematic collection of large-scale data sets (see Hristidis et al. 2010), we also believe that great value might be derived from the systematic collection of interview and survey data that probe and measure coordination across disaster response organizations.

4 Case studies

We present two case studies that are informed by primary data the authors collected immediately after the events and secondary data analyzed by the authors and their collaborators. The authors were on the ground for both events and have access to unique primary data. For Katrina, the case study emphasis is on communication systems infrastructure, whereas for Haiti, the focus is on needs of the affected population. Drawing a parallel between them, we argue that lack of communication prevents organizations from focusing on the right issues. The fidelity of information, security concerns, and novelty of some responders underline opportunities for improvement. Amongst others, one recommendation is to harness the benefits of social media to help humanitarian organizations make the transition to higher information flows. In the years between Katrina and Haiti, the UN established the Cluster Approach, and clusters were in place in Haiti. One of these clusters is the Emergency Telecommunications cluster headed by the WFP. The similarities across these two catastrophes are key to the argument that in both cases, communications and coordination failures hampered the effectiveness of the response. Scholtens et al. (2014) provide a framework of a command and information system as part of a disaster management structure to serve as a control mechanism in information dissemination. They show how the Dutch system failed to meet the requirements of the analytical framework, and how this underscores system failure in practice. In Haiti, the UN clusters were not very effective in the early response (Bhattacharjee and Lossio 2011).

4.1 Sources of data, methods, and data analysis

The first element of Hurricane Katrina data includes the cell tower level repair logs for two major mobile providers, the level of damage of their sites, and the repair actions that they took. The data set covers over three quarters of the installed capacity in New Orleans at the time of the storm. In addition, under the guidance of one of the authors, a group of Master of Public Health (MPH) students at Tulane University conducted post-Katrina interviews. Local emergency managers in St. Bernard and Orleans Parishes were interviewed within 6 weeks of the event. The purpose of the survey was to study the role of information in resource allocation and to determine the sources of information that were used. The study used social network analytic techniques to characterize who were the most important information providers and why. The interviews were conducted across a proportionate distribution of demographics with one interviewer per interview, and the interview transcripts recorded in hand-written notes.

The Haiti case was informed by field interviews conducted from February 2, 2010, through December 2010 combined with a document review of more than 400 primary data sources and an analysis of more than 3400 messages provided through crowdsourcing and registered on the Ushahidi-Haiti crisis map—Ushahidi is an open source mapping technology that uses information from crowdsourcing (Meier 2012; Gao et al. 2011). The technology allows the collection of information for visualization and interactive mapping. Ushahidi, a Swahili word for witness, was initiated after the violence that erupted in the 2007 presidential election in Kenya (Meier and Brodock 2008). Information about the humanitarian crisis and the response that followed was mapped in near real time by volunteers from a variety of sources, including SMS, web, e-mail, radio, phone, Twitter, Facebook, television, lists, live streams, and situation reports. An evaluation of the volunteer effort was commissioned by a student group instrumental in the Ushahidi Haiti Project (UHP) deployment. A “user focus” was requested for the evaluation, and the team conducted a series of preliminary interviews with primary stakeholders to identify an appropriate scope for the evaluation, so it would answer the most pertinent questions for future crisis mapping implementations. Overall, the evaluation was organized around eight main questions, some of which include the following:

• In what new ways did the traditional information systems include the needs of the people affected?

• To what extent do the Ushahidi efforts adhere to norms and standards of humanitarian response?

• Were decisions by responders based on information provided?

• What was the value added by volunteers in classifying and geo-locating the messages?

As part of the Ushahidi evaluation, we interviewed 30 analysts from the UN, the United States Government (USG), and non-governmental organizations (NGOs). These analysts were tasked with capturing, analyzing, and providing decision support for the State Department/USAID, the Defense Department, non-profit organizations, and the United Nations. They typically integrated information from a variety of sources to provide a common operating framework for the decisions and resource allocation of these agencies. The purpose of the interviews was to determine the extent to which this emergent information, especially the Ushahidi crisis map facility, was used and to identify barriers to its use. We found that those who were accustomed to accessing and interpreting unstructured information, such as the intelligence community and defense community, were enthusiastic users of this information in contrast to the traditional field responders, who were unfamiliar with, skeptical of, and often unaware of these sources. The data were largely unstructured, in text format, and came from a variety of sources. In the early stage of emergence immediately after the disaster, before SMS short code became available, data were often remotely registered by individuals who had spoken with affected people on the ground. With SMS, there was a surge in reporting by affected persons; largely requesting survival resources with locations, potential supplies, reports on security, etc. (see Fig. 2). The methodology applied is an iterative, descriptive, and exploratory analysis on the data sets from Katrina and Haiti. We build on prior theoretical drivers of coordination and integration in a disaster. We follow the methodological prescription for the analysis of sociotechnical systems that combines human behavior and organizational context in shaping system behavior (Szajnfarber and Gralla 2017). We adopt the method of collaborative social research by (Miles and Huberman 1994) to collate data into text, field notes, and transcripts (Berg and Lune 2004). We applied the grounded theory approach (Glaser and Strauss 1967) on QSR International’s NVivo 10, a qualitative data analysis program, to explore and code the data sets and extract relevant descriptive and visual statistics from these events. We code analytically and transform them into categorical labels to identify commonalities and isolate meaningful patterns.

Fig. 2

Sources of reports in January 2010. *All non-SMS incidents, after the first few days post-earthquake, were added via the web and thus tagged “Web” submissions. The web category is composed of incidents pulled from e-mail, Twitter and all other non-SMS feeds

4.2 Katrina communication failures

On August 29, 2005, Hurricane Katrina made landfall and impacted more than 90,000 square miles, from Florida to Louisiana. The massive scale of the subsequent flooding exceeded the disaster response plans in the National Response Plan (NRP). According to estimates by the Federal Emergency Management Agency (FEMA), almost half a million people were evacuated (FEMA 2005a), while a million and a half more required financial assistance (FEMA 2005b). With damages of approximately $100 billion, the hurricane had far-reaching socio-economic impacts, and many states across the U.S. declared federal emergencies due to the cost of accommodating displaced people and providing first-responder assistance to disaster areas. More than 7.4 million gallons of oil spilled into the Gulf Coast region’s waterways, the storm crippled thirty-eight 911 call centers, and it knocked out more than 3 million customer phone lines. Broadcast communications were severely affected, with 50% of area radio stations and 44% of area television stations off the air. Compared to other hurricane disasters (such as Camille in 1969, Andrew in 1992, and Ivan in 2004), the federal response, although massive at over 14,000 workers, was widely criticized for its slow response and lack of coordination (Townsend 2006; Dixon 2006). The unexpectedly large failure of the civilian communication network made a dire situation even worse. The Katrina interviews focused on the role of information in resource allocation and on the information sources that were used.

The word cloud, left panel of Fig. 3, shows that communication infrastructure was critical to the massive scale of recovery disruption with “phones and radio channels underlying the needs of relief agencies to manage the situation.” The righthand panel is a summary of the first 24 words, their frequencies and synonyms extracted from the interview notes in response to the question on the major challenges in relief operations.² The responding agencies include the Coast Guard, FEMA, MEMA, TV and radio stations, telecom companies, and relief organizations. The main subject of the discussions includes the hurdles in communication, technological challenges in information transfer and receipt, problems with voice-only systems, etc. over a relative time or period. These discussions were centered on addressing the needs of the agencies, the people affected in the disaster and the services that are required to solve the immediate challenges.

Fig. 3

Katrina Data Analysis: word cloud (left panel); descriptive frequency table (right panel)

As the following excerpts of the responses show, without sufficient information, organizations were unable to focus on the immediate rescue operations or meet the needs of the people: “At first it’s all one-way communication because he doesn’t have time to talk back and forth; he’s just telling people what he needs providing information to those who can provide info.” Another response discusses the absence of guidelines and protocols: “There are no guidelines for communication services—there are no guidelines that talk about pre-deployment, which is what needs to be done to establish prior relationships—only two counties (Hancock and Harrison) had pre-deployed division supervisors for FEMA before the storm. Pre-deployed to build the rapport and get in there before the roads are shut down...” and “Communication have yet to be spelled out in the protocols.”

Another respondent discusses the absence of protocols on access to communication media: “Protocols on who should get phones and who shouldn’t. Just in time training: mechanism on rapid training when you need it. Triage of communication does not exist; it is dependent on ad-hoc. Created a network within 7 days to transmit data.” The following discusses mis-information: “… big problem controlling rumors without the ability of communication within area and outside of area: the lack of means of communication led to isolation; FEMA told police department that they would have a drop of supplies, water and food, to the area.” These excerpts highlight that in the absence of information flows back and forth, it is difficult to know the actual needs of the most vulnerable. This reinforces Marshall McLuhan’s theory on communication (McLuhan 2008; McLuhan and McLuhan 2011). As this excerpt further highlights, the incompleteness of information also implies that the location where the needs were required was not known. The data set from Hurricane Katrina includes the details of the repair logs for two major mobile providers, the amount of damage to their sites, and the repair actions they took (Banipal 2006). Following Hurricane Katrina, commercial mobile and landline communications failed. As many emergency workers also relied on this system, there were additional delays in the efforts to rescue stranded people and direct relief resources where needed. Although many first-responder organizations operate private networks, these systems are not often interoperable, leading to reliance on civilian mobile and landline communications systems to coordinate across organizations. Interoperability across relief agency communication channels is limited because these first-responder organizations have dedicated technologies and broadband encryption that limits cross-communication between the agencies. Though a significant breakdown was anticipated, the long duration and wide coverage area of the service disruption raised concerns about the resiliency of communication networks when they are most needed. Our research indicates that tighter than expected interdependence among different network elements played a decisive role in the failure of communication services at both the civilian and first-responder levels. Figure 4 shows detailed network utilization, cell site availability, and repair log data from two of the leading cellular operators over a period of 3 weeks after the levees failed.

Fig. 4

Leading factors of network outage after the Levees failed. Highlighted date was when over 50 levees and walls protecting New Orleans failed

Analysis of these data shows that 80% of the coverage was disruption due to failure of commercial power and the breakdown of backhaul circuits (which provide communications from cell towers to wired networks). Notice, in Fig. 4, the timeline (in red box) of Katrina occurrence and the levee breaches that resulted in massive flooding. Direct damage to the communication network from wind, rain, and subsequent flooding accounted for the remaining 20 percent of network failure (see Figs. 4, 5). In Fig. 4, generator/battery problems increased after September 2 (yellow bar) and decreased after September 14 because the engineers were steadily repairing the system such that by September 21, only 10% of the system was impacted.

Fig. 5

Impact of Hurricane Katrina on communication network availability

Following the break in the levee, the public switched telephone network was submerged in water, and this disrupted the backhaul network (green bar) that connects each cell tower to the public network and thus severely crippled cellular telephone service.

The second major cause for the disruption of the commercial cellular communication was the failure of the power system. The power system was disrupted, depriving cell site equipment of electricity, and back-up generators were not available at every site. After the hurricane, refueling of generators became an enormous task because of fuel shortages, the breakdown of civil order, rising flood water, and subsequent road closures.

The commercial mobile operators can make their systems more reliable by using microwave antennas to connect network traffic to unflooded zones. In the business districts of major cities, locating cell sites on top of high-rise buildings provides flood protection but exposes sites to high winds. The following interview excerpt captures one approach organizations could take to prepare and coordinate properly given the military’s ability to do so: “Have a back-up radio system and antennas that would be safely stored, and then when the tower gets knocked out, you could get them set up following the storm. Moving up out of the new flood plane would have saved them a lot of problems too; making sure their emergency operations center was not going to get flooded. So extra portable antennas to replace towers that get knocked down,” and “But the military had interoperability; they could access all channels and frequencies.”

4.2.1 Barriers to improvement

Improvements to the civilian communications system may provide social benefits but might be difficult for private profit-oriented firms to justify. For example, during the 2 days prior to the hurricane (during which most of the population evacuated), the telecommunication system experienced an increase in network utilization of over 60% (see Fig. 6). Such an increase over the average utilization rate was responsible for intermittent and infrequent connection of calls (Mattox 2006; Martin 2006). From a business sense, it might be good that a high number of calls are originated on network; however, the costs of providing additional capacity at cell and switch level to handle infrequent spikes in call volumes are significant and the cost of acquiring new spectrum has risen rapidly. The cell sites along the evacuation routes, university campuses, and emergency shelters had the highest peaks. Under normal network traffic conditions, cell sites along the highway (except in cities) carry the least amount of traffic and are usually allocated minimal resources. Only during infrequent evacuations does the network become highly utilized.

Fig. 6

Impact of Hurricane Katrina on Network Utilization

There are two distinct areas of social benefit that might result from increased communication capacity. First, citizens are better able to coordinate during evacuation and immediately after a natural disaster. Families can more easily verify health and safety or reconnect if dispersed. Second, first-responder organizations can utilize civilian networks to coordinate across their boundaries, leading to better allocation of resources to impacted areas. However, profit-motivated firms that make investments to realize returns are unlikely to invest in a system that has adequate resources for the rare but high impact periods of natural disaster. Nonetheless, Rangan et al. (2006) argue, amongst other reasons that private–public partnerships occur in opportunities that require industry-specific competencies with appreciable scale of positive externalities. Implementing Fine’s industry clock speed hypothesis (Fine 1999, 2000) will synchronize how different organizations and even ephemeral networks have access to the improved technology.

4.3 The Haiti earthquake

On January 12, 2010, a devastating earthquake struck the Caribbean island of Haiti. More than two hundred thousand people were killed, and almost 20% of the population was left homeless. As with most other natural disasters, the international community pledged to support rescue and relief efforts. Unfortunately, the rescue efforts were crippled because the devastation was aggravated by the collapse of the nation’s fragile communication network. This made the coordination of emergency response by hundreds of relief organizations and thousands of relief workers difficult and chaotic. Although limited internet bandwidth remained available throughout the crisis due to the dependence of pre-event Internet Service Providers (ISPs) on satellite connections, cell phone providers experienced a loss of cell sites/base stations that exceeded 20% of the nation’s total capacity, with most of the damage near the earthquake epicenter. Except for one station, all radio programming initially went off the air due to destroyed physical and human infrastructure. The already severely limited telecoms capacity was quickly overwhelmed by the extensive use of communications channels in the aftermath of the disaster. But the basic theory postulates that application of these new technologies will result in superior and more real-time information for use in humanitarian response and that responders will use this information to better manage resources that will translate into life- and livelihood-saving humanitarian interventions. Traditional disaster response organizations included teams and non-profit organizations involved in Search and Rescue (SAR) and post-SAR delivery of humanitarian relief interventions. These included UN agencies and a plethora of international non-profit organizations (INGOs) such as Oxfam, CARE, World Vision, and Red Cross. During and immediately following the earthquake in Haiti, local organizations were overwhelmed and trampled by international responders. Overall, the Ushahidi Haiti Project (UHP)³ processed at some level upwards of 40,000 reports. The final posting of UHP reports to the map was the result of the aggregation and processing of an enormous volume of messages originating from a variety of sources (see Fig. 2). Identification of sources, too, was relatively crude due to the rapidly changing volume and sources of UHP reports. In the Ushahidi database, there are 3584 events that have been mapped in Haiti. Of these, 2759 were mapped between January 12 and January 31. Messages were aggregated and classified into main categories as shown in Fig. 7. Consistent with Fig. 6, this summary figure shows that vital lines, categorizing shortage of water, power supply and security, and food constitute the largest percentage of the issues. The reporting on failure of essential services was influenced by the needs for public health and emergency requests.

Fig. 7

Haiti message aggregation and classification

4.3.1 Pareto chart analysis

A Pareto chart analysis, defined as the analysis of the frequency of specific issues or problems or causes in a process, of the categories reveals that most of the report contents are related to essential requirements tagged “Vital lines.” The word cloud on the left panel of Fig. 8 supports the statistics that penurie (food and water), and shortages were overwhelmingly the needs reported.

Fig. 8

Haiti data analysis: word cloud (left panel); summary table (righthand panel)

Interestingly and importantly, ad hoc networks quickly emerged in the hours and days after the Haiti earthquake occurred, giving birth to a skeletal global response network of virtual volunteers around crisis informatics. Internet connectivity was never entirely lost, and radio transmission and enhanced satellite capacity were deployed to the field within the first week of the disaster. Satellite and cell phone companies donated radio spectrum and band-width free of charge.

4.3.2 Emergent networks

The International Telecommunications Union dispatched engineers, 100 satellite terminals, and a Qualcomm Deployable Base Station (QDBS) within a few days of the event. The United States Government focused on rapid communications infrastructure development. The traditional media organizations also deployed equipment to repair the vigorous radio network that existed in Haiti prior to the Earthquake.

The UN system sets up a cluster system of coordination within 2 days of the event. The cluster meetings were to be a place where various data collected through rapid reconnaissance were shared. Even 3 weeks into the response, however, the cluster system was very disjointed and excluded, to a large degree, local Haitian organizations (Haitians had difficulty gaining access to the UN compound, where the meetings were held in English). Very little planning or systematic information gathering was conducted during the period when the emergent information system was vibrant and dynamic. Traditional players relied on more anecdotal information from field officers who may or may not have had a strong vantage point to assess the situation. The key point here is that traditional information sources were virtually absent early after the earthquake because of the degree of destruction and, therefore, the inability of the response/relief agents to conduct rapid assessments or to use other common tools for assessing needs. However, emergent methods provided rapid information organically. Open Street Map is a volunteer group that mapped Port au Prince so that locations were geo-referenced. Networks of virtual volunteers emerged around mobile, geospatial, cellular, and internet technologies, while local affected populations were also incorporated into communications networks to some extent. This emergent network structure was largely organic in nature, though some of the larger donors and information and communications technology (ICT) interest groups promoted its emergence more systematically. An example is the rapid establishment of project 4635, a 911-like SMS short code that was established by technology organizations that were strongly backed by the U.S. State Department. The short code became a tool for affected people to report emergencies and needs. Backend tools were developed that established linkages between affected populations and traditional and emergent responders. The short code was operational by day seven of the crisis, somewhat late to support SAR activities, though there were isolated cases of SAR successes.

Key emergent methods include the establishment and dissemination of SMS short code as a quasi-911 system and the emergence of a successful automated micro-tasking system of translation and management of SMS messages by hundreds of diaspora and local volunteers. At the same time, geospatial analysts in various locations around the world were estimating damage using remote sensing data. This is relevant because traditional actors were isolated in clusters where they retrieved sparse, unsystematic, and not locally informed data. On the other hand, a vibrant organic information system was emerging in near real time, relying on the crowd, e-volunteerism at a scale not experienced previously. This need to provide space for emergent networks reinforces the Galbraith’s theory on boundary spanning (Galbraith 1973).

4.3.3 Geographical information systems (GIS) applications

Another set of efforts, focused on geo-location, attempted to use imagery and other information sources to assign geographic coordinates to distress messages coming from the SMS stream and other information sources. The crisis mapping community attempted to geocode, categorize (put needs according to categories such as basic, vital or essential, medical, etc.) and make critical information available via internet-based maps and direct SMS feeds to responders. Open Street Map is a volunteer project aimed at mapping as many streets and key locations as possible using various sorts of information, including satellite imagery. Ushahidi was deployed as well. Ushahidi comprises organic networks of volunteer groups that linked information from various sources, including the SMS processing activities with human geocoding, text coding, and data base entry to an internet-based map server. The Ushahidi mapping activity was carried out primarily by graduate students from Tuft’s Fletcher School as well as other groups that were recruited by the Fletcher principals (Smith 2010). The Fletcher leader was also involved in the non-profit Ushahidi project, an organization that had grown from an attempt to map and display violence around Kenya’s 2007 election. The ephemeral networks emerged as volunteer cognitive capital was combined with advances in information technologies. The ability to rapidly restore communication by cell phones and maintain internet service allowed students, Haitians, foreigners in Haiti, and new volunteers to become active participants in the remote management of the disaster. A limited number of traditional responders also utilized the emergent information. Lack of information is reflected in several responses, such as, “Has any help reached the Sisters and the people in the soccer field at College STE Marie, Canape Vert?”

4.3.4 Use of information technology by responders

Among traditional responders, only the analyst community within the U.S. Department of Defense and State Department could identify specific decisions that were informed by these data. For example, the Marines identified geographic areas to target their relief activities, and the State Department utilized crisis maps to corroborate their analysis of the relative security of Port au Prince in the early days after the crisis. More often, these analysts used the crisis maps as a situational awareness or sense-making tool, not traceable to specific decisions. Yet analysts interviewed indicated that they consulted the maps daily during the first few weeks after the disaster. Among the most active users of this information were non-traditional responders, which included new organizational players in the relief operation, such as small and new non-profit organizations and individual citizens who wanted to help. An example of the latter is the owner of a marketing firm in Canada, who connected with a Haitian American youth in Miami by Twitter to launch a joint distribution of relief resources in the Haitian youth’s village of origin, retrieving relief resources from the Port au Prince airport before any of the traditional responders were actively delivering relief. When asked what prompted her to engage, as a first-time responder, she said simply “because I can, and the technologies make this possible.” On the other hand, except for the use of initially processed SMS messages during the SAR phase of the response, emergent information networks were not used by most traditional responders in the field.

One senior staff member of a large NGO summed it up well: “big players will never use information from these unreliable sources.” A U.S. Government official explained, “so much is done by standard operating procedure; staying in your comfort zone.” The reluctance to incorporate information from a variety of sources poses a distinct risk that humanitarian responders will not be sufficiently responsive to save lives. A key issue is distrust of the global crowd, considered to be too inexperienced to provide information with sufficient fidelity. Similarly, traditional responders typically distrusted the accuracy of messages from affected Haitians.

4.3.5 Lessons learned

The Haiti case shows how telecommunications infrastructure can be rapidly restored after a disaster and combined with new technologies, especially social networking tools, to enable effective responses. For example, this response from our survey data reflects how social media can assist in timely response: “Please help find him: (name withheld) sent a text message stating that he is stuck underneath the rubble at his workplace... Unfortunately, I do not have a photograph of him. Any info, write me via Facebook, Twitter.” This shows that information, when transmitted, often expose relief needs; however, the outcome of the ensuing response efforts is largely unknown, creating a feedback gap. Several factors limited the use of these emergent information efforts by the broader humanitarian community. First, the traditional responder community expressed concerns about the fidelity of information and the quality of information processing. The involvement of untrained human volunteers was cited as a great disadvantage.

Second, the new data streams were largely unstructured in nature, and traditional responders, used to traditional indicators, did not know how to use the new information. Third, the traditional community identified well-founded security concerns that included publishing information that identifies the location, age, and gender of potentially vulnerable individuals. Boundary conflicts between new and traditional responders are other barriers to the use and adoption of these new information tools. Traditional responders took a very defensive posture vis-a-vis the criticisms of the Haiti response as slow, bureaucratic, and expensive. On the other hand, the new volunteerism brought to bear the resources of youthful energy and cognitive surplus immediately, flexibly, and with very little funding.

The cases illustrate that the chaos following catastrophes can be alleviated by creative and novel configurations of agents who are increasingly responding remotely (on a global scale) as well as local resident populations who have traditionally been hindered as responders by communication network failures and exclusion from traditional responder information systems. The cases show the following:

• 1. Cellular networks became more resilient to catastrophe between Katrina and Haiti. This is a result of technological evolution with the emergence of more integrated cell phone and internet services. Thus, the information fidelity in the tripod of Fig. 1 is reinforced.

• 2. Internet and mobile computing tools have created a space for non-traditional volunteers and organizations to become significant actors in an emergent ephemeral response net-work. This network is now global in character and novel in its ability to mobilize cognitive capital and resources at a global scale. This ties back to the data and investment in the tripod presented in Fig. 1.

• 3. Organizational boundary spanning, and conflicts remain significant obstacles to an effective model of disaster response, giving rise to two distinct needs in theory development. The first is rapid closed loop information flows. The cases demonstrate that information, when transmitted, often exposed relief needs but not the outcome of response efforts. This lack of feedback hinders rapid learning during events. The second need is to enable systems to make faster decisions even in an unstructured environment, thereby reducing negative outcomes. The emergence of unstructured social media tools as a powerful response force suggests the possibility of improving the ability to operate in an unstructured environment. Both large and small-scale data sets will be critical to allow researchers to calibrate more formal analytic and decision support models—see the tripod in Fig. 1 on feedback. We provide a sample interview guide in the “Appendix” as an example of the types of data that could be collected.

Finally, efforts to span boundaries between disparate organizational actors occur with varying scales of organization. The chaos of catastrophe can now be matched on an even greater scale with the chaos of response. Reducing this chaos requires effective linkage through information flows. We identify that systems evolve independently to optimize performance along dimensions that are most critical to each. Thus, we present the observational framework that building better interfaces to existing standards is important so that as the systems evolve, users are still confident of interoperability. In summary, improvement of any of the tripod in Fig. 1—feedback, data and investment, and information fidelity—is required to promote resilience.

5 Discussion

The response to disasters by organizations shows a repetitive pattern, where organizations learn in leaps, i.e., disaster by disaster, rather than smoothly over time (Carley and Harrald 1997) and where retention of knowledge is highly unpredictable. Thus, a model of organizational learning (inter- and intra-organizational) that is dynamic and promotes un-freezing. Unfreezing is the initial phase of a change process whereby the motivation to change is created (Schein 1972). In addition, we highlight the values of interoperability, coordination, and interdependence of different systems.

5.1 Organization design

A major challenge is to ensure that the overall responder industry has a carefully thought-out plan to maintain information flows across organizational boundaries even after the organizations evolve. This is currently being done—through periodic responder drills, through the frequent coordination meetings that take place between the major first responders; through an emphasis on interoperability in the field—and must be improved over time. To help understand how to manage boundary spanning, we prescribe Galbraith’s (1973) framework in which intra-organizational capacity to process information can be increased either by the creation of lateral relations or through investments in information systems.

5.2 Interoperability

While interoperability of information systems is a recurrent theme in disaster management literature, overemphasis on these systems reduces the connectivity of emergent civilian non-traditional actors who are heavily reliant on open source and emergent social and mobile computing tools, such as Twitter, Facebook, Ushahidi/crisis mapping, and related tools. More deliberate planning for the incorporation of emergent tools needs to be incorporated into exercises and preparedness activities, and traditional actors should be trained in the use of these techniques. More systematic training and support of digital volunteers will provide greater fidelity of information. Memoranda of understanding (MOUs) between organizational providers of cognitive surplus and the traditional community might better harness these volunteers to address needs that other actors are less able to handle, such as data mining (to close information loops and extract learning in motion), systematic archiving of big data, and locating response resources outside of traditional supply chains.

5.3 Coordination

Coordination failure is one of the most important contributors to failed efforts such that co-location is commonly applied in the form of the Emergency Operations Center. Nevertheless, the incorporation of individuals who are dedicated to boundary spanning using old and new information management strategies has not yet systematically emerged and should be considered as a deliberate strategy in the establishment of emergency operations center (EOCs). An EOC is a central command and control location that is tasked with the execution of emergency preparedness and the management of protocols for response after a disaster. Similarly, as part of preparedness capability development, these types of information boundary spanners should be incorporated into organizations that are active in disaster management to systematically improve the information culture of disaster response organizations.

5.4 System interdependence

It is observed from the Haiti case that the weakness of the connections between relief organizations is exacerbated by what appears to be a lack of creativity. The different capabilities of analysts and non-analysts have in their information use can lead to conflict. For example, poor response performance during the Haiti crisis was not due to cascading failures, as in other disasters, but rather it was partly due to the novelty of the players where traditional organizations were less agile in their data use than the military or non-traditional organizations. Thus, we propose that relief tasks must be segmented by degrees of humanitarian experience—humanitarian experienced or not experienced.

To wrap up, the major hurdle here is that of implementation. How can we ensure that we can have sustainable continuity? We draw lessons from how the military is organized for disasters (Weeks 2007; Gaydos and Luz 1994). First, the military has a central collection point for information exchange. Second, the military uses organizational maps. This will allow ephemeral organizations to be familiar in structure to improve efficiency and sense-making. Third, the military covers all the bases; that is, in the military, the staff structure contains all the requisite skills required in a disaster. Fourth, barriers to action are removed such that hierarchical layers or boundaries are bypassed in an emergency.

6 Conclusion

The absence of information flows among disaster response organizations often leads to poor operational performance. To limit response failures, it is imperative to institute policies that mitigate the collapse of communication infrastructure when disasters occur (Sellnow et al. 2002). We develop two cases that illustrate how using evolving technology and information flow to facilitate disaster management can be viewed as a coordination problem inspired by communication failures. We find that internet and mobile computing tools have created a space for non-traditional volunteers and organizations to become agents or significant actors in an emergent ephemeral response network that provides cognitive capital and resources on a global scale. The analysis of Haiti data shows that the SMS came online too late for its actual use as a type of 911 tool. Most of the earlier information came from internet sources and largely outside of country. Within the country, early information was rudimentary, such that the types of needs, food and water, reflect needs several days into the disaster as opposed to SAR needs, which is what the SMS would have been particularly helpful for.

Breaking the bureaucratic hurdles faced by traditional first responders is a prerequisite to enabling spontaneous and ad hoc responses. Given the limitations on how resources flow, we highlight that providing information to emerging support mechanisms allows them to provide useful responses. A growing number of responders, such as members of religious organizations and citizen groups, often lack access to resources. To enable these actors, budgets should be allocated not only to existing groups, but to emerging groups to break barriers created by bureaucratic silos; however, there are still obstacles to an effective model that seamlessly integrates the different response components. For example, barriers to organizational boundary spanning and inter-organizational conflict during disaster response prevent the rapid closed-loop information flows that enable rapid learning during events. We use the case studies to illustrate the evolution towards information-driven interdependent systems. Our suggestions for moving forward include ICT infrastructure as a social good that requires subsidy/investment, increased emphasis on boundary spanning, a stronger emphasis on archiving and utilizing big data, and more rapid integration of emergent methods and crowd techniques. While the cases, a hurricane and an earthquake, are different, it is evident that the underlying issues for a successful relief operation, particularly with respect to resilience in communication—are invariant to the disaster type. We posit that improving the resilience of communication may very well be a one-size fits all strategy to disaster preparedness because communication and coordination are central to effective efforts. We conclude by highlighting that the three phases of enhancing and managing resilience as articulated ERMG project require continuous investment (Antunes et al. 2017).

Appendix

1.1 Disaster relief agent/administrator interview guide

In this “Appendix,” we provide an example template for capturing key information from agents or organizations during or after a disaster.

Responder background during your career, have you ever worked or been involved in any major disasters? If yes, what roles did you play? Did you participate as an individual or with a relief agency, a government entity, or as a member of a faith-based organization?

Nature of the disaster describe the nature of the disaster: was it a hurricane, tsunami, earthquake, flooding, industrial accident, or terror attack? How were you involved?

Relationship of your job to the relief operation Does your regular day job involve any emergency response or first-aid response? If not, how did you acquire the skills to enable the response you provided?

Added knowledge Using this experience, describe how your spontaneous reactions or contributions to the relief align with that of the relief organization that you joined or worked for. How were your activities coordinated? What provided the mechanism for different responders’ activities to be integrated?

Operational challenges what challenges did you face in your role as a first responder? Did any prior knowledge come in handy? If you had no prior knowledge, what challenges did you face? Can you give specific examples?

Boundary spanning challenges what challenges did you observe as different organizations coalesce to coordinate for a common goal? Did you face any boundary spanning problems due to the different operational modes or approaches from different response agencies? How were these differences addressed? Can you give examples? Probe: did language or cultural differences (especially with international agencies) present any obstacles? Provide stories on specific incidents: need stories on jargon and procedures)?

Coordination challenges When organizations have conflicts on approach or strategy, how were they resolved? Please give an example or two. Which mechanisms seem most effective in resolving conflicts? (How do you measure effectiveness?) Probe (only if interviewed after survey): Do you ever use the following: standards, joint meetings, co-location (in which direction?), shared information systems, dedicated personnel? Probe: if there were shared information systems with other organizations or agencies, what are the systems employed, and are they standard or modified. Probe: what has your experience been with using standards? Could you give some examples?

Other personnel What kinds of experiences or background help make you or other people effective at relief coordination? Please give an example or two. What sort of training does your agency or organization offer? What was your experience with the “soft skills” training, if any?

Curriculum If we were to design management program that focused on relief supply chain management and integration, what types of skills (or courses) would you suggest that universities concentrate on? Please give some examples of why you would include these skills or courses.