Living through Organizational Chaos: The Emergence of Order from Technology-Based Organizing 1 ABSTRACT

Transcription

1 Living through Organizational Chaos: The Emergence of Order from Technology-Based Organizing 1 Ning Nan Price College of Business University of Oklahoma Yong Lu Information Sciences and Technology Penn State Hazleton ABSTRACT In this study we seek to understand how social order emerged from a technology-based organizing process in the aftermath of a tremendously destructive natural disaster. Our research context is a type of social medium (a campus online forum) at a university 50 miles away from the 8.0-magnitude earthquake that struck southwest China in By premising the sociomaterial perspective, our analytical focus shifts away from the impact of discrete technology entities or human behaviors and toward the process of organizing actions that are jointly enacted by both humans and technology. We apply the complex adaptive systems model of information technology (IT) use (Nan 2011) to the analysis of 27,271 online forum posts in order to assemble a precise and move-by-move account of the key properties and mechanisms involved in the recurrent entanglement between humans and technology during the organizing process. The findings indicate that humans and technology are mutually-constituting forces in an organization s ability to resume order amidst an unexpected disruption of normal organizational life. This study enables us to rethink the relationship between technology and organizing in several important ways. Keywords: organizational chaos, organizational order, sociomateriality, complex adaptive systems, technology. 1 Manuscript under review at MIS Quarterly. 1

2 Living through Organizational Chaos: The Emergence of Order from Technology-Based Organizing INTRODUCTION A 8.0-magnitude earthquake, thousands of aftershocks, an extended campus shutdown, frequent dorm evacuations, and losses of friends and families: when these events all happened to over 22,000 students at a university merely 50 miles away from the earthquake s epicenter, casual intuition led to the expectation of panic and chaos. However, in the aftermath of the devastating earthquake in Wenchuan, China on May 12, 2008 these students enacted an incredibly orderly campus life. They lined up patiently for food and water, and willingly followed instructions from the school s central administration. Although evacuations and camping conditions opened the door to exploitative behaviors, witnesses observed only cooperation and assistance. By the end of the first week after the earthquake these students even organized several disaster relief activities. Intrigued by this orderly episode in the chaotic aftermath of a tremendously destructive natural disaster, we examined this university s campus online forum a type of social medium and a rich repository of human cognitions, emotions, actions, and technology appropriations both before and after the earthquake. Although massive earthquakes rarely occur, organization members do face situations that require them to make sense of and enact order into unexpected, unknown, threatening, or fearful disruptions of normal organizational life (Quinn and Worline, 2008). These situations have been identified as valuable opportunities for expanding the boundaries of scholarly thinking since they often reveal properties and causal pathways that are less visible in more tranquil periods (Meyer, 1982; Weick, 1988, 1993). We correspondingly believe that an examination of the campus online forum immediately before and after this massive earthquake would allow us to rethink organizing and information technology (used interchangeably with technology hereafter) in 2

3 two important ways. First, as one of the very few functioning communication channels in the aftermath of the earthquake, the campus online forum was deeply engaged in the organizing actions triggered by this dramatic disruption of normal campus life. Supplied with a complete record of the online forum, we have the opportunity to provide a full account of how technology and humans collaboratively produced social order. We are motivated to seek novel theories and analytical approaches best suited for this effort since the traditional human-centered or technocentric analysis of technology and organizing does not fully capture the integral and active role of technology in organizational life (Orlikowski 2007). Second, Weick et al. (2005, p 409) pointed out that, organizing itself is embodied in written and spoken texts ; the written texts in the campus online forum therefore present a longitudinal view of a technology-based organizing process. By analyzing the written texts posted at different points of time (e.g., before and after the earthquake), we can gain a move-by-move perspective of how social order is produced and reproduced by technology-based organizing actions over time. Compared with previous information systems (IS) studies that portray organizing activities as technologically-occasioned political dynamics and organizational changes (e.g., Avgerou and McGrath 2007; Levine and Rossmoore 1995; Silva and Hirschheim 2007), this move-by-move perspective can more precisely represent the constantly-evolving and never-ending entanglement between technology and humans in everyday organizing (Leonardi and Barley 2008; Orlikowski 2007). The objective of this study is therefore to apply the novel sociomaterial perspective (Orlikowski and Scott, 2008; Pickering, 2008) and the complex adaptive systems model of IT use (Nan, 2011) toward rethinking the relationship between technology and organizing. In the context of the aftermath of a massive earthquake our rethinking is guided by three research questions: what are the human responses to a dramatic disruption of organizational life? What 3

4 are the technology appropriations in response to the disruption of organizational life? And most importantly, how do these human responses and technology appropriations collaboratively produce social order over time? We employ the sociomaterial perspective in order to shift our analytical focus away from discrete technological or social factors and toward the shaping power of the entanglement between technology and humans. Our answers to these research questions therefore directly engage the integral and active role of technology in organizing practices. The complex adaptive systems (CAS) model of IT use serves as an IS-specific middle range theory for integrating previous fragmented insights regarding technology and organizing into a coherent research model. This CAS research model allows us to identify the key properties of human responses and technology appropriations, as well as the mechanisms whereby these human and technology properties jointly produce an orderly life within a chaotic organizational context. The sociomaterial perspective in concert with the CAS research model provides an analytical advantage for distilling a full and move-by-move account of the inseparable role of technology and humans in the resilience of today s organizations from 27,271 university online forum posts. In the remainder of this article we first establish a novel theoretical lens for this study by reviewing and rethinking the theories and research concerning technology and organizing. We then apply this novel theoretical lens to an empirical analysis of this university s online forum in order to assemble a full and move-by-move depiction of the technology-based organizing process. Finally, we present the implications and conclusions of this study. RETHINKING TECHNOLOGY AND ORGANIZING Prior Research Given our research interest in the technology-based organizing actions during a period of unexpected disruption of normal organizational life, our theoretical development began with a 4

5 review of the previous research on the relationship between technology and organization as well as on organizing actions in unexpected organizational contexts. An assessment of the progress and gaps in the prior research enables us to specify and justify the point of departure for our rethinking of technology and organizing. Prior research has generally recognized technology and humans as two primary sources of variations in the relationship between technology and organizing (Leonardi, 2011; Markus and Robey 1988). Building on the assumption that technology and humans have distinct roles in organizations, prior research has typically chosen to focus on the impact of either technology or humans. This leads to two well-recognized lines of work: techno-centric and human-centered (Orlikowski, 2007). Since several studies have provided thorough accounts of these two lines of work (Leonardi and Barley, 2008; Markus and Robey, 1988; Orlikowski, 2007), our discussion only highlights the key differences between these prior theoretical conceptions and analytical approaches (see Table 1 for an overview). These theoretical and analytical differences are a major cause of a persistent research gap in the IS and organizational literatures as detailed below. Table 1: Overview of Prior Research Techno-Centric Human-Centered IS Research Organizational Research Analytical Focus Impact of technology Influence of humans Influence of humans Conception of IT Exogenous forces Consequences Contexts (assumed away) Level of Analysis Organizational Individual Individual Primary Modeling Variance-based Process-based Process-based Approach Central Thesis Organizational form and function are determined or constrained by technology Example Studies Bjorn-Andersen et al., 1986; Carter, 1984; Kling 1980; Robey 1977; Pinsonneault and Kraemer, 1993; Stewart, 1971 Human cognitions and actions give meaning to interpretively flexible technology Barley, 1990; Orlikowski et al., 1995; Yates and Orlikowski, 1992; Majchrzak et al., 2000 Sensemaking of human actors enacts order into unexpected disruptions of organizational life Hutchins, 1991; Gephart, 1993; Quinn and Worline, 2008; Meyer, 1982; Weick, 1988, 1993; Weick et al.,

6 The central thesis of the techno-centric research stream is that organizational form and function are either determined or strongly constrained by technology (Markus and Robey, 1988). Although referred to as techno-centric, this line of research typically revolves around technology-related organizational changes such as the centralization of authority (e.g., Robey, 1977), middle management downsizing (e.g., Pinsonneault and Kraemer, 1993), and the routinization of work (e.g., Kling, 1978). Conceived as an exogenous force, technology is often reduced to a nominal (e.g., presence or absence of a technology) or surrogate measurement (e.g., the amount of technology investment). This research stream employs variance-based models as the dominant analytical approach in evidencing a significant association between technology and its organizational impacts. While techno-centric research focuses on the impacts of technology, the human-centered research stream privileges the shaping power of human actors. In the IS literature the humancentered approach is represented by research on the social construction of technology (see Leonardi and Barley 2010 for a thorough review). A central theme of this social construction research is that technology is interpretively flexible and that its functions and effects are defined by goal-oriented human cognitions and actions. With the exception of the variance-based examinations of user perceptions (e.g., Fulk et al., 1987; Griffith and Northcraft, 1996), social construction research has primarily employed a process-based analytical approach for tracing the content and influences of individual behaviors in technologically occasioned events (e.g., technology design, adoption, or uses). A stream of human-centered research particularly relevant to this study from the organizational literature is the work on organizing in unexpected organizational contexts. The central concept of this organizational research is conveyed by the word sensemaking (this 6

7 research is therefore called organizational sensemaking research). Sensemaking refers to the sequence of mental activities and organizing actions triggered by the discrepancy between an individual s expected and experienced states of reality; order is enacted into ongoing circumstances as human actors recurrently extract cues and retrospectively make plausible sense of events (Weick et al., 2005). As in human-centered IS research, organizational sensemaking research has primarily relied on the process-based approach in order to depict and examine the influences of human actors in unexpected organizational contexts such as a terrorist attack (Quinn and Worline, 2008), a medical emergency (Weick et al., 2005), a labor strike (Meyer, 1982), or a work accident (Weick, 1993). Technology is effectively absent from the organizational sensemaking research, presumably because it is viewed as a contextual factor. While the techno-centric and human-centered literatures have evidenced the relevance of technology and humans respectively, their divergent theoretical conceptions and analytical approaches do not illuminate how technology and humans jointly produce social order in today s increasingly digitized organizations. A growing number of studies have shown that although technology and humans exist independently of each other, their roles in organizing practices are a function of how they are enacted together (Zammuto et al., 2007). Leonardi (2011, p.149) pointed out that, [b]y themselves neither human nor material agencies are empirically important. But when they become imbricated interlocked in particular sequences they together produce, sustain, or change either routines or technologies. This fundamental conflict between the separation of technology and humans in extant research against the inextricable roles of technology and humans in everyday organizing practices has caused a long-standing paucity of research that theorizes and examines technology as integral to organizational life (Akhlaghpour 7

8 et al. 2009; Leonardi and Barley 2010; Orlikowski and Iacono, 2001; Orlikowski and Scott 2008; Zammuto et al. 2007; see Appendix A for a summary of previous reviews). This persistent research gap defines our theoretical point of departure from the existing literature. In order to achieve a richer understanding of the integral and active role of technology in organizations, we require a new vision of the inseparable roles of technology and humans in organizations. We also require a new research model that can reconcile the different conceptions and analytical approaches regarding the relationship between technology and organizing. Our theoretical development of the complex adaptive systems model with the sociomaterial perspective is therefore directed at addressing both requirements. A New Research Focus: Sociomaterial Entanglement Building on the ontological vision where, both the human and nonhuman are recognized as open-endedly becoming, taking on emergent forms in an intrinsically temporal dance of agency (Pickering 2008, p. 1), a sociomaterial perspective has been proposed to embrace the equally important and tightly intertwined roles of technology and humans in today s organizational life (Orlikowski, 2007; Orlikowski and Scott, 2008; Suchman, 2007). A growing body of sociomaterial theoretical expressions and empirical studies are gradually shifting our analytical focus in two important ways (we summarize the sociomaterial literature in Appendix B for a concise presentation). First, the sociomaterial perspective redefines the unit of analysis for IS and organizational research. By highlighting the inseparable role of technology and humans in organizations, the sociomaterial perspective justifies a move away from discrete technology entities or human actions and toward a temporal flow of organizing practices jointly produced by both technology and humans. Suchman (2007, p. 267) pointed out that, [the sociomaterial studies] respecify 8

9 sociomaterial agency from a capacity intrinsic to singular actors to an effect of practices that are multiply distributed and contingently enacted. Therefore, we can gain considerable analytical insight if we give up on treating the social and the material as distinct and largely independent spheres of organizational life [and replace] the idea of materiality as pre-formed substances with that of performed relations (Orlikowski, 2007, p. 1438). Corroborating these theoretical arguments, recent empirical studies have demonstrated the analytical advantage of this new unit of analysis in generating rich and robust evidence for the integral role of technology in organizations. For example, Wagner et al. (2010) shifted their analytical focus from either the best practices encoded in technology or established practices in user communities alone to the recursive negotiations and accommodations between the two during a package enterprise system s post-roll-out period. This new analytical approach revealed the active role of technology in turning around a troubled system in recursively defining and being defined by the negotiated practices. Similarly, an analysis of the sequence of synergistic interactions between technology and humans (called imbrication ) during the development and usage of a simulation tool offered a compelling account of how the material agency of technology (i.e., technology s capacity to act on its own) served as a shared building-block in organizational routines and technological infrastructures (Leonardi, 2011). Second, in addition to redefining the unit of analysis, the sociomaterial perspective changes our view of the causal pathways intrinsic to the relationship between technology and organizing. Social order becomes the aggregate property of the temporal flow of organizing practices rather than the direct impact of technology or humans via the sociomaterial lens. This view is indicated in the assertions that, the pattern of open-ended extension through modeling, dialectics of resistance and accommodations, and so on is the only explanation that I can 9

10 defend of what scientific culture becomes at any moment, (Pickering, 1995, p. 147), as well as statements that, the point in the end is not to assign agency either to persons or to things but to identify the materialization of subjects, objects, and the relations between them as an effect of ongoing sociomaterial practices. (Suchman 2007, p. 286) By using the sociomaterial perspective we can direct our research effort away from testing unidirectional relationships between technology and humans, and toward analyzing how the recurrent intertwining between technology and humans at the individual level gives rise to social order at the organizational level. This new perspective of causal pathways is evident from field observations that system survival is the organizational-level outcome of the recurrent negotiations between best practices and established practices on the individual level (Wagner et al., 2010). Leonardi s (2011) ethnographic study also depicts changes in organizational routines and technological infrastructures as the aggregate history of the synergistic interactions between technology s material agency and humans goal-oriented actions (Leonardi, 2011). The sociomaterial perspective therefore shifts our analytical focus away from the static effects of discrete technological entities and human actions, and toward the dynamic process whereby social order arises from organizing practices produced collaboratively by technology and humans in an organizational context. An examination of this dynamic process transcends the conventional separation of technology and humans in the IS and organizational literatures. It can therefore close the research gap concerning the integral and active role of technology in today s organizational life. This new analytical focus has been referred to as shifting assemblages (Orlikowski and Scott 2008), the mangle (Pickering 1995), ongoing sociomaterial practices (Suchman 2007), the imbrication (Leonardi 2011), and negotiated practices (Wagner et al., 2010). In order to streamline the terminology we call it sociomaterial 10

11 entanglement in this article. Our study of the technology-based organizing in the aftermath of a massive earthquake is therefore focused on sociomaterial entanglement. A New Research Model: Complex Adaptive Systems While the sociomaterial perspective shifts our analytical focus toward the inseparable roles of technology and humans in organizations, it does not prescribe an analytically precise and consistent approach for encoding empirical observations of sociomaterial entanglement into a formal research model. We therefore require an IS-specific middle range theory that defines a temporal stream of organizing actions as its unit of analysis, conceives social order as the aggregated property of these organizing actions, and more importantly specifies the key properties and mechanisms involved in the process whereby social order arises from these organizing actions over time. The complex adaptive systems (CAS) model of IT use (Nan, 2011) is identified as an ideal theory. The CAS model of IT use has recently been developed for capturing the dynamic process whereby organizational-level IT use patterns and outcomes arise from individual-level IT use behaviors (i.e., the bottom-up IT use process). Although the term sociomaterial entanglement is not invoked in this model, the bottom-up IT use process in effect manifests the hallmark of our study s analytical focus: individual-level IT use behaviors embody the organizing practices collaboratively produced by technology and humans while organizational-level IT use patterns and outcomes capture the social order arising from organizing practices. The elements of the CAS model outlined below serve to codify the key properties and mechanisms concerning sociomaterial entanglement in a precise and holistic way (see Table 2 for an overview). Drawing on the tenets of CAS theory (Gell-Mann 1994; Holland, 1995; Miller and Page 2007), the CAS model of IT use is formulated around the three core elements of every complex 11

12 adaptive system: agents, interactions, and an environment. Agents are individual actors or basic entities of actions in a CAS; they are described by attributes and behavioral rules. Attributes are the internal states of agents (Epstein and Axtell, 1996). Behavioral rules are the schemata governing an agent s attributes or behaviors; they can be considered as a set of input or output statements linking an agent s perception of the world to changes in the agent s attributes or actions (Drazin and Sandelands, 1992; Epstein and Axtell, 1996; Holland, 1995). Table 2: A Coherent Research Model of Sociomaterial Entanglement CAS IT Use Conception in this Study Agent Human Actors Individual users involved in sensemaking actions in online forum. IT Features Technological components of the campus online forum involved in sensemaking actions. Attribute Individual Differences Discrepancy between users expected and experienced states of reality. Technology Interpretive flexibility of IT features. Characteristics Behavioral Rule Mental Activities Users cognitions and emotions activated by the discrepancy between expected and experienced states of reality. IT Functionalities Adaptive capacity of IT features. Interaction User-System Mutually adaptive behaviors between users and IT features. Interactions Interpersonal Interactions Mutually influencing behaviors between users. Connection User-System Links Submission of posts to the online forum. Interpersonal Ties Communication links between users in the online forum. Flow Movement of Intangible IT Resources Codification and distribution of knowledge, information, attitudes, beliefs, or other intangible resources via the online forum. Environment Environment The unexpected disruption of normal campus life. Structure Social and Organizational Structures Macroscopic Observations Social Order Context for the university campus such as instructions from the school administration. The temporal pattern of aggregate properties of human actors, IT features, user-system interactions, interpersonal interactions, and organizational environment as embodied by the campus online forum. Building on prior IS and organizational research, the CAS model of IT use recognizes human actors and IT features as the two basic categories of agents (see Table 2). In reference to the analytical focus of this study, human actors and IT features represent the two building-blocks (i.e., humans and technology) of sociomaterial entanglement. The attributes of human actors and 12

13 IT features are defined respectively as individual differences (e.g., demographic backgrounds and cognitive states) and technology characteristics (e.g., flexibility and reliability). The behavioral rules of human actors are embodied by mental activities associated with cognitions and emotions. These mental activities link a human actor s perception of the world to conscious or unconscious changes in internal states and behaviors. IT functionalities represent the behavioral rules of IT features. As the capabilities for IT features to act on their own (i.e., the material agencies of technology, Leonardi 2011), IT functionalities determine how IT characteristics and behaviors are either preserved or modified as a consequence of interacting with human actors. This conception of attributes and behavioral rules in the CAS model of IT use captures the empirical specifications of humans and technology in sociomaterial entanglement. Interactions refer to agents mutually adaptive behaviors (see Table 2). As the most commonly observed structures in a CAS (Drazin and Sandelands, 1992), interactions embody the temporal stream of organizing actions or recurrent intertwining of technology and humans in organizations. They can be viewed as a function of agents, connections, and flows. Agents are the processors of interactions; agents behavioral rules contain the logic for converting patterns in received information into changes in the agent s attributes and behaviors. Recurrent applications of behavioral rules produce a temporal stream of interaction patterns. Connections are relational links among agents as delimited by the attributes of agents; they designate possible channels where interactions can take place. Flows refer to the movements of resources such as information through the web of agents and connections, and serve to inject energy that sustains agents interactions (Holland, 1995). 13

14 The CAS model of IT use recognizes user-systems interactions and interpersonal interactions as the two primary ways for mutually-adaptive behaviors to unfold. User-systems interactions involve human actors and IT features as agents, and direct uses of technology as connections. In contrast, interpersonal interactions are enacted by human actors and carried out via interpersonal ties such as friendship or joint membership in communities. The exchange, codification, and reuse of information, knowledge, belief, or other intangible resources via direct uses of technology or interpersonal ties embody user-system or interpersonal interaction flows. By explicating interactions via agent, connection, and flow concepts, the CAS model provides a concrete description of how technology and humans collaboratively produce the temporal stream of organizing practices (i.e., interactions). These user-system and interpersonal interactions meld the empirically distinguishable specifications of technology and humans into their conceptually inseparable role in organizations. An environment is the medium agents operate within and interact with (Epstein and Axtell, 1996), defined by structures that characterize the topography of an environment. In the CAS model of IT use, environmental structures are represented by organizational contexts such as cultures and rules. Agents, interactions, and environment together produce macroscopic observations of technology use patterns and outcomes. These macroscopic observations represent the social order arising from the temporal stream of organizing practices collaboratively created by technology and humans. An important insight from the CAS model is that the social order, being the logical consequence of the attributes and behavioral rules of human actors and IT features, cannot be reduced to the direct effects of attributes or behavioral rules (Nan, 2011). This is consistent with the essence of the sociomaterial perspective: neither technology nor humans are 14

15 empirically important by themselves (Leonardi 2011). Instead, they take on emergent forms in an intrinsically temporal dance of agency (Pickering 2008). The CAS model of IT use provides an overarching framework for unifying the different conceptions and approaches of prior research into a coherent research model. This allows us to examine sociomaterial entanglement as embodied by the campus online forum. Recall that techno-centric research is primarily concerned with the organizational impacts of technology using variance-based models. This research in effect provides a static view of sociomaterial entanglement s macroscopic observations (i.e., the social order) via the CAS lens. The dependent variables in its variance-based models are snapshots of agents aggregate properties (e.g., the number of middle managers), interactions (e.g., centralization of authority), or the environment (e.g., routinization of work). By extending the static view of techno-centric research toward the dynamic process of sociomaterial entanglement, we define social order as the temporal pattern created from the aggregate properties of all human actors, IT features, usersystem interactions, interpersonal interactions, and the organizational environment as embodied by the campus online forum. By focusing on microscopic observations of sociomaterial entanglement, human-centered research provides important implications regarding the attributes, behavioral rules, connections, flows, and the environment in the CAS model. In particular, the social construction of technology research suggests that the most important attribute and behavioral rule of IT features are indicated respectively by their interpretive flexibility and adaptive capacity. Technology s interpretive flexibility is the ease for human actors to modify the material makeup of IT features (Leonardi, 2011). It is this open-ended nature of technology that enables the dance of agency between technology and humans. Adaptive capacity references IT features inscribed abilities to 15

16 change alongside the goal-oriented human actions. It fundamentally determines how IT features are preserved or modified via sociomaterial entanglement. The organizational sensemaking research complements the social construction of technology by specifying the meaning of goal-oriented human actions. During an unexpected disruption of normal organizational life, the central goal of human actions is to make sense of and enact order onto ongoing circumstances (Weick et al., 2005). In reference to this goal, a key attribute of human actors is cognitive state regarding the discrepancy between expected and experienced states of reality. The mental activities (cognitions and emotions) triggered by this discrepancy become in effect human actors (agents ) behavioral rules. These behavioral rules subsequently generate sensemaking actions in the form of reading, writing, and editing the online forum s content (Gioia et al. 1994). Within the boundary of these sensemaking actions, usersystem links are represented by submitting posts to the online forum (i.e., initiating or extending a discussion topic), interpersonal ties are embodied by communication links among the users who submitted these posts (i.e., one user responds to another user s post), flows refer to the distribution of information, knowledge, attitudes, beliefs, or other intangible resources via the online forum, and the environment is the unexpectedly disrupted normal campus life. We formulate a coherent research model of sociomaterial entanglement in an unexpected organizational context (refer back to Table 2) by integrating the fragmented insights gathered from both techno-centric and human-centered research. This coherent research model enabled us to synthesize our empirical findings regarding the human responses and technology appropriations triggered by the earthquake into a full and move-by-move account of a technology-based organizing process embodied by the content of the campus online forum. 16

17 METHOD Data Collection The context of our study was the 8.0-magnitude earthquake that struck Wenchuan, Sichuan Province in China on May 12, It was named, [the] most devastating earthquake since the founding of new China [in 1949]. (Zhou, 2008). Over 69,000 people were confirmed dead, 17,900 missing, and 374,000 injured as a consequence of this earthquake (USGS, 2011). We chose a major public university (pseudonamed FEU for the remainder of this article) located 50 miles away from Wenchuan as our data collection site for two reasons. First, since FEU is not located in a region with frequent earthquakes its 22,000 students had no a priori earthquake response training or emergency plans. More importantly, due to its geographic proximity to the epicenter FEU s normal campus life was substantially disrupted by the earthquake (e.g., classes were cancelled and all nonessential services were shut down). These circumstances ensured that the earthquake was perceived as an unexpected disruption by FEU students, and this disruption was sufficient to trigger the organizing actions required for re-establishing orderly campus life. Second, the campus online forum was the primary channel for students divided between FEU s two campuses (10 miles apart) to communicate and organize during the earthquake s aftermath. Normally students only had access to the forum between 8 AM and 11PM ( lights-out was from 11PM to 8 AM; note that most universities in China cut off electricity supply to the dorms during night hours). During the first week following the earthquake FEU s central administration decided to provide electricity 24 hours a day so that students could use the online forum whenever they desired. This decision fostered the technology-based organizing embodied by the campus online forum, and allowed the forum to capture a rich record of the FEU students responses and technology appropriations during the period immediately following the earthquake. 17

18 We retrieved all 27,271 posts submitted during the first week following the earthquake (May 12 to 19, 2008) directly from the online forum s server. For each post we collected data regarding its topic, content, the user ID of its submitter, the date and time it was submitted, and the sub-forum where it was listed. In order to understand the social organizational context of these posts we also collected concurrent data from official updates on the number of aftershocks, earthquake-related casualties, and formal announcements issued by FEU s central administration. Data Analysis Our analytical approach can be generally described as an interpretive case study grounded in the CAS research model (refer back to Table 2). Two steps were taken in order to analyze the 27,271 posts retrieved from the campus online forum s server. First, we developed quantitative indicators for the CAS elements regarding the users, forum features, and interactions (see Table 3). These quantitative indicators lent precision to our interpretation of the case, enabling us to determine the granularity of our move-by-move account of the technology-based organizing process according to the changing patterns in these quantitative indicators. Among the quantitative indicators we directly calculated the number of users submitting posts each day (the indicator for users) and the number of posts submitted (the indicator for usersystem links) each day. The development of the other quantitative indicators involved three methods: source code analysis, social network analysis, and open coding. Source code analysis was used for obtaining the indicators of forum features. Whenever a user utilized a feature beyond plain text a source code was embedded in the post. For example, the inclusion of a URL in a post would be indicated by the source code [url] embedded within the post s content. An examination of the embedded source codes within the collected posts revealed four features: 18

19 quotes, URLs, attachments, and emotion icons. The quantitative indicator of each forum feature was therefore defined as the number of posts containing the feature per day. Table 3: Overview of Quantitative Indicators of the CAS Research Model CAS Elements Quantitative Indicators Human actors: individual users involved in Number of users submitting posts each day. sensemaking actions on the online forum. Individual difference: discrepancy between users Percentage of earthquake-related posts among all expected and experienced states of reality. posts submitted each day. Mental activities: users cognitions and emotions Number of posts showing negative emotions, activated by the discrepancy between expected and information, appreciation, self-reflection, action, and experienced states of reality. emotional support respectively each day. Technology features: basic components of the campus online forum involved in sensemaking actions. Characteristics: interpretive flexibility. Functionality: adaptive capacity. User-system interactions: the mutually adaptive behaviors between users and forum features. User-system links: submission of posts to the online forum. Movement of resources: the distribution of knowledge, information, attitudes, beliefs, or other intangible resources via the online forum. Interpersonal interactions: mutually-influencing user behaviors. Interpersonal ties: communication links between posts Movement of resources: distribution of knowledge, information, attitudes, beliefs, or other intangible resources via the online forum. Number of posts containing the source codes respectively for quotes, URLs, attachments, or emotion icons each day. Number of posts showing respectively voting and social networking each day. The same indicators as those for users, user-system links, and movement of resources (as a function of these three elements). The number of posts submitted each day. The same indicators as those for mental activities. The same indicators as those for users, communication links, and movement of resources (as a function of these three elements). The number and degree of centralization for communication links each day. The same indicators as those for mental activities. Social network analysis was employed in order to generate quantitative indicators for communication links. Based on previous research (Ahuja et al., 2003; Wasko et al., 2009), a communication link was defined as the incidence between two users when one replied to another s post. We created an adjacency matrix for each day. In this matrix users in each row were those who initiated discussions (the first post in a discussion thread), and users in each column were those who replied to others posts (subsequent posts in a discussion thread). We counted the number of links and calculated the total degree of network centralization. Network 19

20 centralization indicates the extent to which communication links were concentrated around one or a few users. This is measured in comparison to the most centralized network a star network where a central user connects to all other users, while all other users are only connected to the central user (Wasserman and Faust, 1994). Organizational Risk Analyzer (Carley et al., 2007) and UCINET (Borgatti et al, 2002) were used in order to analyze and visualize these social network indicators. Table 4: Coding Scheme CAS Coding Elements Categories Discrepancy Earthquakerelated content Mental Negative activities, or emotions movement of resources Information Interpretive flexibility, or adaptive capacity Action Descriptions Posts concerned with the earthquake or its aftermath Posts expressing fear, fright, anxiety, sadness, anger, tiredness, or frustration Posts seeking, providing, or synthesizing information Posts suggesting, planning, or giving support to actions Example Posts [5/13/08, 7:29 AM. Who did you think of first during the earthquake?] I thought of a girl I like. [5/13/08, 6:41 AM. Reply: It s dawn.] I am too frightened to sleep. I fear that I will never wake up. When will this end? [5/15/08, 8:15 AM. Will the blood donation bus come to campus?] I heard about the bus on TV. Does anyone know whether and when will the bus come to FEU? [5/16/08, 6:41 PM. Reply: Disaster relief.] How about organizing a freshman team to tutor high school students in the worst-hit regions? [5/14/08, 9:06 AM. Reply: To those in heaven.] My deepest sympathies! Let s give a helping hand!!! Emotional Posts expressing sympathy and support compassion Appreciation Posts praising others [5/15/08, 12:27 AM. The most awesome people.] Those at the campus radio station are still working at this hour. You are the best! Self-reflection Social networking Voting Note: these categories are not mutually exclusive. Posts evaluating or criticizing collective identities or actions Posts trying to establish or maintain one-to-one contact Deliberate replies or discouragement of replies to moderate the ranking of a post [5/15/08, 2:37 PM. Reply: Are we inconsiderate?] I don t agree. At least my class took the trash with us after camping on the open ground. [5/18/08, 6:40 PM. Reply: Activity for spreading love flying blue ribbon.] Qiao [a user s name], have you received a blue ribbon? [5/16/08, 7:46 PM. Reply: Clothing on sale! Proceeds will be donated!] Ding We used an open coding approach (Corbin and Straus 1990) in order to obtain the remainder quantitative indicators (i.e., discrepancy, mental activities, movement of resources, interpretive flexibility, and adaptive capacity). This approach did not invoke an a priori coding scheme; instead, it allowed categories to emerge from the data via an iterative and multiple-step interpretation process. First, we identified posts related to the CAS elements in question from all the earthquake-related posts. Next, the first author went through several iterations of comparing 20

21 and grouping posts in order to generate a comprehensive set of categories for the corresponding CAS elements (see Table 4). The same categories were identified for mental activities and the movement of resources because the content of these posts reflects both elements. Similarly, interpretive flexibility and adaptive capacity shared the same two coding categories since the identified technology appropriations could indicate both elements. After all categories were reviewed, revised, and agreed on by the coauthors, a coder trained in content coding but unfamiliar with our research questions was hired to identify the posts falling into each category. A second coder was then hired to evaluate a randomly-selected subset (10%) of the posts. Both coders were provided with the coding scheme presented in Table 4; the inter-rater agreement for the two coders was 89.1%. The second step of data analysis was the theory-driven interpretation (Walsham, 1995). Armed with the knowledge of the changing quantitative indicator patterns, we sought to understand the meanings embedded within the technology-based organizing process (Orlikowski and Baroudi, 1991); we employed verbal analysis in order to explain why sociomaterial entanglement unfolded the way it did at our research site. The CAS research model provided a precise vocabulary for explaining the key properties and mechanisms embedded within the human responses, technology appropriations, and social order jointly produced by humans and technology. These two data analysis steps produced a full, precise, and move-by-move account of technology-based organizing during an unexpected disruption of an organizational life. RESULT The resulting quantitative indicator values (see Table 5) showed substantial changes in the CAS elements regarding the users, forum features, and interactions during the first week 21

22 following the earthquake. For example, users mental activities were initially dominated by negative emotions, and then transitioned to informational and action-related activities (see Figure 1); emotion icons were the most frequently used forum feature during the first five days, but voting, quotes, and URLs later caught up (see Figure 2). Table 5: Results of the Quantitative Indicators Quantitative Indicators 5/12 5/13 5/14 5/15 5/16 5/17 5/18 5/19 U: Number of users D: Percentage of earthquake-related posts M: Number of posts with negative emotions M: Number of posts with emotional support M: Number of informational posts M: Number of action-related posts M: Number of appreciative posts M: Number of self-reflective posts F: Number of posts including quotes F: Number of post including URLs F: Number of posts including attachments F: Number of post including emotion icons F: Number of posts using social networking F: Number of post with voting F: Number of posts C: Total number of links C: Total degree of centralization U: users; D: discrepancy; M: mental activities, movement of resources; F: flexibility, adaptive capacity; C: communication links N u m b e r o f P o s t s Information Action Negative emotion Emotional support Appreciation Self reflection N u m b e r o f P o s t s Social networking Voting Quote URL Attachment Emotion icon Date Date Figure 1: Users mental activities or movement of resources. Figure 2: Forum features interpretive flexibility or adaptive capacity. The changing patterns in the quantitative indicator values suggest four general moves in technology-based organizing: the accumulation of negative emotions immediately following the earthquake; transitions to informational, appreciative, or self-reflective mental activities; construction of actions; and eventually the re-establishment of a normal life. In the remainder of 22

23 this section we will present the results of our interpretation of the technology-based organizing process along these four general moves. For each move we will first describe our observations from the online forum, and then use the CAS research model as a consistent theoretical lens to explain the key properties and mechanisms involved in this move of the technology-based organizing process. Move 1: Earthquake as an Unexpected Reality (May 12 13, 2008) Before the earthquake hit at 2:28 PM on May 12, 2008, the FEU campus online forum was business as usual. From 8:00 AM (the daily opening time of the forum) users began to submit posts either as new discussion topics or as replies to an ongoing discussion thread. The discussion threads were organized into 36 sub-forums. Most of these sub-forums were created per user requests after the implementation of the forum system. Each sub-forum was recognized by a distinct theme such as movies, music, schoolwork, or sports. Each contained userdeveloped discussion threads indicative of its theme (e.g., Tips for Job Interviews in the Job Hunting sub-forum). When replies in a thread began to drift away from its topic a user would often post a comment saying, building tilts ( building is the slang for a discussion thread). This helped the threads to stay on topic. Users employed the forum features that fit the content of their posts (e.g., a post about photography would use the attachment feature to show photos). Each user selectively replied to discussion threads according to his or her interests. The top-ten most replied to discussion topics across all sub-forums were displayed on the front page of the forum system. Immediately before the earthquake the top ten topics included About Appearance, New Moderator for the Cartoon Sub-Form, and Favorite Soft Rock Songs. None of these posts were concerned with any current or potential disruption of campus life. 23

24 These pre-earthquake observations provide a benchmark for the sociomaterial entanglement situated on the online forum. The key properties of this benchmark can be described in terms of the CAS research model: when users perceived minimal discrepancy between their expected and experienced states of reality, then their personal interests were the main attributes that delimited selective uses of forum features and discretionary communication links with other users. The continuous operation of the online forum was maintained by the energy injected into the forum as knowledge, information, beliefs, and attitudes regarding users personal interests were codified as forum posts and shared via communication links (i.e., flows); the recurrent applications of users interest-driven mental activities and forum features adaptive capabilities to these mental activities produced predictable collective behaviors (i.e., the social order) on the campus online forum (e.g., all posts regarding the same personal interest were selforganized into a sub-forum). The most interesting aspect of the pre-earthquake sociomaterial entanglement is that neither the key properties of forum users nor those of forum features showed a strong tendency to seek order, yet they created the social order on the campus online forum without the intervention of FEU s school administration 2. Previous CAS studies suggest that this is possible when there are positive feedback loops among agents behavioral rules (Anderson, 1999; Drazin and Sandelands, 1992; Resnick, 1994). Positive feedback loops arise when the application of one agent s behavioral rules is reciprocated by other agents applications of the same behavioral rules. Over time these positive feedback loops can reinforce some behavioral rules and diminish others, leading to stable structures on the collective level. On FEU s online forum these positive feedback loops unfolded in two ways: when a user s post regarding a specific personal interest 2 The online forum system s hardware was FEU s property, while the FEU Student Association established and managed the form s daily operations. 24

25 was replied to by other users posts discussing the same interest, and when forum features adapted to interest-driven user behaviors by affording the possibility of threads and sub-forums. These mutually reinforcing feedback loops between humans and technology are the key mechanism that produced the online forum s stable structures. Any deviation from these stable structures was quickly corrected without the intervention of a central controller (e.g., the building tilts warning to a strayed post). Table 6: Discussion Thread of Strong Earthquake Hit Campus Time Post 2:37 PM, 5/12 My heart is still pounding; during the earthquake life felt so fragile thank Heaven I am alive! 2:39 PM, 5/12 Reply 1: It is an earthquake, no kidding humans are so trivial, oh, we are so lucky! 2: 54 PM, 5/12 Reply 2: The computer in the next room was shaken to the floor. 2: 55 PM, 5/12 Reply 3: Wow, you were still online during the earthquake? We ran out of the dorm; some people didn t even have clothes on. 4: 41 PM, 5/12 Reply 4: A number of guys ran out of the dorm in their underwear. Buildings have cracks. Things collapsed to the ground. A lot of people are standing on the open ground. 4:54 PM, 5/12 Reply 5: I was reading on the fifth floor in the library. The whole building was rocking, everyone rushed down, people ran in an orderly fashion, my heart is still racing. 7:34 PM, 5/12 Reply 6: The earthquake lasted a long time. It kept shaking I was in the Sunshine Square. I heard it was really bad on the sixth floor. I wonder how those in Liulin [one of the two campuses] are doing, I can t get hold of any of them. 7:43 PM, 5/12 Reply 7: I was alone in the dorm during the tremor. I ve never felt that lonely, two minutes felt like several hours. I am lucky; so far 58 people have been confirmed dead in Chengdu :`( My condolences. 8:32 PM, 5/12 Reply 8: It was truly frightening. I felt like the sky was falling down and I had nowhere to flee 8:38 PM, 5/12 Reply 9: Just got a text message saying there may be aftershocks tonight. Be safe!! 9:08 PM, 5/12 Reply 10: I am finally back to the dorm :`( :`( :`( 9:24 PM, 5/12 Reply 11: How will we survive tonight!!? 9:36 PM, 5/12 Reply 12: Very shaky, earthquake for sure, I saw things collapse to the ground. The dorm is a mess. This is horrible. 10:41 PM, 5/12 Reply 13: The earthquake [also] hit Liuling [a nearby county], but it wasn t very bad. 11:13 PM, 5/12 Reply 14: I am reading this thread in the midst of another aftershock. 11:23 PM, 5/12 Reply 15: This feels like a movie, no one will sleep tonight. 11:27 PM, 5/12 Reply 16: My legs are still shaking, an earthquake ~so frightening! 12:06 AM, 5/13 Reply 17: I escaped death, no fear no fear no fear. 1:05 AM, 5/13 Reply 18: Several aftershocks just hit, but they were mild. 1:59 AM, 5/13 Reply 19: Another aftershock. 2:00 AM, 5/13 Reply 20: I want to cry. 2:03 AM, 5/13 Reply 21: How will we live through tomorrow? 3:51 AM, 5/13 Reply 22: Does anyone know the situation in Beichuan [a nearby county]? I hope my family there is ok. 12:02 AM, 5/14 Reply 23: I also worry about Beichuan. I heard it was hit really bad there :`( devastated ~they lost all communications with the outside. 25

26 In the early afternoon of May 12, 2008, it was quiet on both FEU campuses since students were either in classes or in the dorm taking a break. The majority of FEU students had never experienced a massive earthquake. We can therefore understand the shock when the 8.0- magnitude earthquake rippled through both FEU campuses at 2:28 PM. Ten minutes after the earthquake the first earthquake-related thread was initiated on the online forum (see Table 6 for the entire thread). Not surprisingly, it expressed each user s shock at the earthquake. The replies submitted during the next six hours were personal reports of what one did (e.g., ran out of the dorm, Reply 3), saw (e.g., things collapse to the ground, Reply 4), or felt (e.g., the sky was falling down, Reply 8) at the moment of the earthquake. In addition to these first-hand personal accounts, a few earthquake-related threads used the URL feature to link to news reports of the earthquake. Since none of the sub-forums had a theme related to unexpected events, this and other earthquake-related threads were developed on a variety of available sub-forums such as Flea Market and Job Hunting. Moreover, replies to several non-earthquake-related threads began to tilt toward earthquake, but no user tried to correct this by pointing out building tilts. The level of perceived discrepancy and negative emotions increased with the accumulation of first-hand accounts and second-hand news reports of the earthquake on the forum (refer back to Figure 1). As a user expressed, the more I read these reports, the more frightened I am. During the evening statements such as I want to cry, I am dying, and I am paranoid became increasingly prevalent. The emotion icon for crying [ :`( ] was a frequently-used forum feature (e.g., Replies 10 and 20 in Table 6). In the evening of May 12, FEU s central administration issued a formal announcement that all classes had been cancelled until May 14 for safety reasons and only essential services would be available 3. Some users expressed a feeling of being ignored and abandoned by the university. During the first night 3 Classes and nonessential services actually resumed on May 26, 2008 due to persistent aftershocks. 26

27 following the earthquake FEU students reacted strongly to several mild aftershocks. According to personal accounts submitted to the forum, many FEU students rushed out of the dorm whenever there was a slight tremor. A few users attempted to persuade others to stay calm; however, their posts were typically discredited by replies such as, it s suicide to stay put and why don t you respect your own life? Figure 3: Communication links before 2:28 PM on 5/12/2008 Figure 4: Communication links between 2:29 PM 5/12/2008 to 11:59 PM 5/13/2008 We compared the pre-earthquake benchmark and sociomaterial entanglement through the first night of the earthquake in order to identify the key properties of human responses and technology appropriations during the first move of the technology-based organizing. The differences are evident: following a dramatic event in the environment (i.e., the earthquake) salient user attributes changed from personal interests to the discrepancy between expected and experienced states of reality (the percentage of earthquake-related posts rose from 22% on 5/12 to 77% on 5/13, see Table 5). Driven by this shared cognitive state of discrepancy, users appropriated the forum features in ways that facilitated their expressions of the shocks (e.g., uses of emotion icons increased from 14 on 5/12 to 250 on 5/13, see Table 5); their communication links became more centralized around earthquake-related discussion topics (degree of centralization rose from on 5/12 to on 5/13, see Figure 3 and 4 for a visual comparison). The resources flowing through these communication links were primarily negative emotions such as fear and anxiety (the number of posts with negative emotions rose from 52 on 27