Business processes as sociomaterial networks ... - Semantic Scholar

5 downloads 0 Views 262KB Size Report
based quality management project in a large aircraft maintenance ... business process modeling, within a large aircraft maintenance ..... aircraft crashes somewhere? .... Tampa, Florida, 2003. [3] Chang, J.F. .... of Social Theory, 5(2), 2002, pp.
Business processes as sociomaterial networks: exploring the multiple dimensions of flexibility in process modeling João Porto de Albuquerque Department of Computer Systems, ICMC-USP University of Sao Paulo, Brazil [email protected]

Marcel Christ Department of Informatics University of Hamburg, Germany [email protected]

Abstract

studies claim that BPM is currently in its third generation (the so-called Third Wave) [2, 3]. This new generation is attempting to bring together the benefits of past BPM approaches by seeking to model and manage the processes, while simultaneously maintaining the capacity of the organization for change and thus its ability to readily adapt to new and unforeseen situations. Thus, the goal of flexibility has clearly been given prominence in recent approaches to process modeling. Process modeling thus currently pursues two different objectives: the documentation of work practices into diagrams and flexibility. These two objectives may appear to be contradictory: process modeling entails formalizing processes through models, so that work can be structured and organizational activities made more uniform and predictable. At the same time, the documentation of organizational practices in models, and ensuring these models are strictly adhered to, can reduce the degree of organizational flexibility. This is because the fixed nature of the process activities that result can prevent an organization from responding to new and unforeseen situations. For this reason, process modeling projects must tread a fine line between the formalization of processes (possibly leading to automation) and adopting strategies to preserve organizational flexibility. This ambivalent ambition of process modeling is bound up with the wider theoretical investigation into the relationship between documents/models and organizational practices. The following question arises: what are the implications of process modeling for organizational flexibility? Research on business process management has often adopted a technological approach, particularly with regard to flexibility [4]. As a result, flexibility is mainly regarded as an attribute of the technical artifacts which have to be achieved by means of an appropriate configuration or the employment of specific techniques in that artifact. Although these approaches are useful, they fall short of addressing

In the last few years, it has become common for organizations to document work practices by means of business processes diagrams. An analysis of the assumptions underlying process modeling, shows that this procedure pursues two seemingly contradictory goals: on the one hand, it aims at formalizing work practices into diagrams that set out a defined sequencing of activities; on the other, it seeks to bestow flexibility on the organization – i.e. to allow it to maintain its ability to respond to unforeseen situations. On the basis of this seemingly contradictory motivation, this paper analyzes the implications of process modeling for flexibility by adopting a sociomaterial perspective that is grounded on the Actor-Network Theory. This is employed to conduct an empirical study of a processbased quality management project in a large aircraft maintenance company. The main contribution of this study is to provide a sociomaterial, multidimensional interpretation of the relationship between process modeling and flexibility.

1. Introduction In recent decades, the concept of the business process has increased in popularity and become widespread as a basic construct for structuring organizational work in conjunction with information systems in the so-called Business Process Management (BPM) [1, 2]. One of the core activities in BPM is the modeling of business processes, i.e. the documentation of work practices in the form of process diagrams or models, which are generally based on a graphical notation like that of a flowchart. These diagrams define the activities of a process and their mutual relations, in a way that allows the established activity sequencing to be subject to restrictions that must be observed in social practice. Although the practice of documenting work with flowchart diagrams is not a new phenomenon, recent

the complex social and organizational negotiations that take place in process modeling projects, as well as how the flexibility is (and can be) achieved in practice. In other words, empirical studies are needed that provide a deeper understanding of modeling projects and the way they can achieve (or fail to achieve) flexibility [5]. This paper seeks to make an additional contribution to the existing empirical studies on business process modeling [6, 7], particularly by tackling the issue of flexibility from a sociomaterial perspective. In the pursuit of this goal, we adopt an approach which draws on the Actor-Network Theory [8, 9]. This sociomaterial perspective is used in the present paper for conducting an empirical study of an integrated quality management system, based on business process modeling, within a large aircraft maintenance corporation. The remainder of this paper is structured as follows: Section 2 outlines the theoretical underpinnings of the research; Section 3 describes the research design and methodology employed. Following this, the case study is described and analyzed in Section 4, and Section 5 concludes the paper by holding a discussion of the results obtained.

2. Formal Models, Organizational Practices and Flexibility In analyzing the relationship between business process models and organizational flexibility, account should be taken of how these models are related to organizational work practices. Business process models are often conceptualized as portrayals or representations of patterns of action in the organization, called business or organizational processes. Although they have a distinctive connotation that is derived from the “process view” mentioned earlier, processes can be basically seen as organizational routines. In fact, although the terms “process” [10] and “routines” [11] are used in the literature in an ambivalent way, both are generally associated with recurring patterns of action that are embedded and anchored in the organization. Recent research studies have highlighted the dynamic character of organizational routines [12-14]: since they are human practices, they must be carried out in a continuous way to allow routine activities (or processes) to be undertaken and in this way make it possible for variations and changes to occur. In fact, the relation between formal models and work practices has been the focus of a large number of research studies during the last few years [15]. As Berg [16] points out, it is commonly assumed that

there are essential differences between “dead” formal artifacts – which are usually regarded as abstract, homogeneous, and symbolic – and “live” human actions – whose messy, dynamic and contingent character can never be entirely mapped into formal artifacts. If this differentiation is transposed to the organizational domain, it is necessary to distinguish between formal, “dead” business process models and the represented informal, “live” business process [14]. In this line of reasoning, there is an underlying assumption that formal artifacts (e.g. process models) and enacted organizational practices (e.g. business processes) belong to distinct and detached domains of reality: whereas the former are affiliated to the technical realm of mechanical determinism, the latter form a part of the living world, and are the locus of human agency and interpretation. Although this dichotomy of two ontological categories tends to be implicitly assumed rather than explicitly theorized, it has practical consequences. From this standpoint, organizational flexibility must be regarded as an exclusive attribute of the social, human domain of work practices [13]. Since models are elements of the formal world, they are impoverished representations of these social practices. This means that models are rigid by their very nature and hence, formalization/control and flexibility must be seen as mutually exclusive objectives, which are frequently represented as antithetical poles of a one-dimensional axis [17]. As a result, there is a trade-off between formalization and flexibility: Each step in the direction of more documentation/formalization means a corresponding loss of flexibility. In recent decades, academic research in the field of Science and Technology Studies (STS) has challenged the putative ontological dichotomy that exists between the social and material (or technical) world, and it has been argued that this essentialist view is not able to account for the complexity of the situation found in practice, since human activity is closely intertwined with technical and formal artifacts [9, 18, 19]. In fact, the reification of “essential properties” in both formal artifacts and human activities conceals some of the implicit and necessary counterparts from the “other side”. These include, for instance, the “configured” human behavior that is implied in the workings of the artifacts [20], as well as the constitutive role played by material artifacts in social relations [9]. Moreover, as a result of the entanglement of formal artifacts and human action, “new competences for workers can be achieved, higher levels of complexity in work tasks can be reached, and activities can be coordinated over time and place”

[16]. Recent scholarship in the field of organization studies has echoed these notions and, it has been argued that before the relationship between technology and organizational practices can be fully understood, we must question the validity of the ontological dichotomy between the “social” and “material” and recognize the relevance of materiality in organizational practices [21-23]. Furthermore, as the empirical research on the coconstitution of documents and organizational contexts has shown, documents “do not simply force certain reading on their reader” [24]. As pointed out by Suchman [19], situated actions that emerge from the embedding of a plan (or model) in social practices cannot be regarded as a mere ‘execution’ of the plan, since these actions have their own dynamics and those who carry them out must be prepared to deal with the contingencies of new situations. By analogy, it would be misleading to consider that process models (such as documents) “force” the performance of certain work practices, and that this will inexorably lead to a loss of flexibility. Building upon the insights found in these studies, we argue that on the basis of the ontological premise that underlines the essential dichotomy between “live” processes (on the social side) and “dead” artifacts (on the material side), it is impossible to grasp the constitutive character and generative power that business process models may have in practice. In a way that is analogous to the findings of Lee and Hassard [25], we argue that the essentialist divide that attributes flexibility solely to the versatility of humans, is not a useful method of analyzing how (and to what extent) the liveliness and flexibility of organizations operate in practice when business process models are used – especially now that modelers are explicitly incorporating flexibility as a goal (as seen in the previous section).

2.1. Processes networks

as

sociomaterial

actor-

To overcome the ontological divide outlined above, we draw on theories derived from Science and Technology Studies and known as the Actor-Network Theory (ANT). ANT is the most prominent body of literature within “sociomateriality” studies [23]. As units of analysis, ANT relies on heterogeneous, sociomaterial arrangements called hybrids [26], or actor-networks [9, 27]. These entities comprise both human and non-human ‘actants’ – a word borrowed from semiotics to refer indistinctively to human and non-human actors, such as people, texts, concepts, machines and so forth.

The association of the different actors to build a relatively stable network of alliances is known in ANT as ‘translation’, i.e. a process that “relates, defines, and orders objects, human and otherwise” [28]. It should be underlined however, that the relative stability of the relations that are translated and “black-boxed” in a network is not guaranteed, but should be seen as a precarious [27] and uncertain [9] achievement that needs to be continuously enacted in organizational practices. Thus, an actor cannot be said to exist “outside” of the network, because it simultaneously constitutes and is constituted by the network of relations with the other actors. This means that the terms actor, actornetwork, and actant can, in effect, be employed interchangeably (as they are in this work). Following Latour [9, 29], we understand ANT as a relational ontology or metaphysics, in which the entities of the world (actors) are constitutively defined by the set of relations they perform with other entities (the network). In view of this, ANT can be seen as a part of the larger body of practice theories [30] that rely on a relational thinking where “subjects, social groups, networks, or even artifacts develop their properties only in relation to other subjects, social groups, or networks” [31]. The relational ontology of ANT should not be confused with the proposition of complete equivalence between humans and non-humans, but as a refusal to take for granted a priori divisions between material/technical and social features [23, 32]. Thus, ANT provides a sociomaterial lens that is capable of overcoming the ontological divide explained in the previous section by revealing the sociomaterial actor-networks within which the organizational practices and business process models are co-constituted. From the sociomaterial perspective of ANT, the formalization of organizational practices that takes place in business process modeling can be seen as the knitting together of a heterogeneous, sociomaterial actor-network around a business process model. In other words, a model artifact (i.e. an actor) is created together with a relatively stable set of relations around (i.e. the network), and is put into circulation. This entails a departure from viewing process modeling as an abstraction in its most common connotation, i.e. the production of impoverished representations of organizational practices that are caused by the subtraction of most aspects of organizational reality. Instead of this, the sociomaterial perspective of ANT regards process modeling as the addition of a new actor-network to the organizational scene, i.e. as the creation of a model artifact that simultaneously builds a

sociomaterial network around it. Moreover, this perspective entails regarding model artifacts and their putative properties (e.g. rigidity, formality, durability etc.) as ongoing achievements in the performances of the network relations. From this perspective, before it is possible to consider how flexibility can be understood and whether it can be achieved in practice (or not), we must examine the hybrid, sociomaterial actornetworks within which the organizational routines and business process models are co-constituted.

2.2. The irreversibility of networks and flexibility The concept of irreversibility of translations as defined by Callon [8] is instrumental in investigating the flexibility of the heterogeneous networks that are built around models of business processes. The degree of irreversibility of a network is said to depend on two things: “(a) the extent to which it is subsequently impossible to go back to a point where that translation was only one amongst others”; and (b) “the extent to which it shapes and determines subsequent translations” [8]. This means that, the irreversibility of a network inhibits its ability to change, i.e. it is antagonistic to the concept of flexibility. As Callon points out, this definition of a translation is a relational matter, which “can only be measured when it is put to the test” [8]. Thus, before one is able to investigate flexibility in process modeling, it is first necessary to identify the heterogeneous actor-networks that are built around the process models, and map their constituent elements. Following this, it must be determined whether the arrangement of the network elements engenders a state of irreversibility in the translations and thus leads to inflexibility. This task can be accomplished by examining how process networks are able to deal with competing translations, i.e. how they cope with changes that challenge the relations and associations established in the network. The concept of irreversibility is particularly useful in this context because, as it is based on the premise that a process model is constitutively embedded in a heterogeneous actor-network, it implies that flexibility is a property that arises from the sociomaterial network as a whole. Hence, flexibility cannot be analyzed by only taking account of the properties of the isolated constituent elements of the actor-networks involved, but consists of an emergent property (in the sense that is used in System Theory) from those sociomaterial networks.

3. Research Design and Methodology This paper seeks to understand the implications of process models on the flexibility of organizations, i.e. their ability to change. This question is analyzed by means of a case study that involved a process modeling project led by the quality management department of a large-scale aircraft maintenance company in Germany (henceforth called AMC). We adopt a qualitative, interpretive approach as a guiding principle [33]. The empirical material was obtained by means of interviews about the historical background, and the reasons for undertaking the project. It also included direct observations carried out in several modeling workshops, and an analysis of secondary sources. In summary, the sources of information for this research include the following: a) semi-structured interviews with members of the quality management department, including the head of the project and two process modelers; b) field notes taken during the observations in some of the modeling workshops and tutorials about the developed software tool; c) e-mail interviews with the process modelers, to address questions that came up during the analysis; d) secondary documents, i.e. printed documents and manuals for instructing employees about new features and how to use the software tool, official Internet information about the company, and the official environmental statement of AMC. The data was collected by one of the authors over a period of about four years (2007-2010), as a part of his larger PhD project. While the data was being collected, the analysis was conducted in a collaborative and interactive way to provide guidance on other issues that had to be addressed during the fieldwork that followed [34]. The analysis followed the procedure proposed by Latour [9] of “follow[ing] the actors themselves or rather that which makes them act, namely the circulating entities”, which is analogous to the principle of the hermeneutic circle of Klein and Myers [34]. In our case, this entailed examining the empirical data and attempting to distinguish between the different actor-networks that were constituted during the process modeling workshops of the project. In the initial stage, this involved coding the data by making use of the elementary concepts actor, network, change, irreversibility, and flexibility. This enabled us to identify the main actors and networks in the analyzed project, as well as to disclose different conceptualizations of flexibility from the perspective of the organizational actors themselves. The next stage consisted of providing a description of the main actor-networks identified, and establishing a

connection with the different kinds of flexibility. This led us to abstract and generalize by employing the concept of “dimensions of flexibility” explained in Section 4.3. Finally, the proposed dimensions of flexibility were checked with the interviewees by conducting further interviews and e-mail exchanges, with a view to improving the internal validity of the study.

4. Case Analysis AMC has about 22,000 employees and is a global network with affiliates and subsidiaries spread around the world. According to the head of the quality management department, AMC's Business Process Modeling (BPM) project which is analyzed below was triggered by a technical failure in a plane during its takeoff. It was this incident that led the quality management department to decide to reorganize and re-establish its quality management system. Until then, quality management had been regarded by people in AMC as simply an inconvenient requirement of the aeronautical authority. In fact, the company had a repository of interrelated documents such as operational instructions, safety measures, and environmental guidelines, as well as quality management directives. These were designed to ensure the maintenance of standards at work, but basically consisted of lengthy documents and a wide range of heterogeneous diagrams. These documents were drawn up by different business departments, each using its own language and particular definitions of procedures and roles. As a result, before it was possible to find out about business processes in different business units, it was necessary to read a large number of documents with various graphical notations, different formats, and divergent terminologies. Thus, it is not surprising that these documents were rarely used.

4.1 Prelude: the Quality Management Project When we began this study, the Quality Management Project (QMP) had been carried out on a national level for about five years. It was aimed at improving the quality of standards and making business processes more transparent and easier to manage by gradually modeling them for each business unit. The ultimate goal was to integrate all the organizational processes into a corporation-wide, easy-to-use process map that could be used by employees to visualize the process models – thus

getting rid of the old text-based documents and nonstandardized diagrams. Although the focus of our analysis lies in the modeling processes (explored more fully below), it is useful to start by briefly analyzing the establishment of the project network. In our case, the quality management department must interact with the following actors (as seen from its own perspective): Business Unit Managers: these are concerned with improving the quality of the business processes in their units. They are also assumed to be responsible for asking the employees in their units to become involved in the project. Employees: these have a practical (tacit) knowledge of business processes, but their practices are uncoordinated and sometimes in conflict. They tend to adopt a critical stance to quality management, since it may entail additional work. However, they are interested in tools that can make it easier for them to carry out their activities. Modeling Framework/Tool: this enables the processes to be standardized and provides a better form of visualization by documenting the work practices as business process diagrams. Aeronautical Authority: this seeks transparency in providing it with access to the process models so that it can check that the rules and regulations are being complied with. In this manner, the project sets out by establishing a system of alliances between the actors involved, which in turn (re)defines their interests and groups them around a common goal. However, the project will only be effective if the modeling is successful and the process models are used in practice – and this leads us to the question of the modeling workshops.

4.2 Modeling workshops: defining processes and owners The modeling of business processes was carried out in workshops and conducted by a modeler from the quality management department with up to ten (generally three to six) employees as representatives of their business units, e.g. repair personnel (on the shop floor), lawyers (legal department) and sales staff (sales and insurance departments). The representatives were interviewed by the modeler about the most important activities carried out, decisions made, and data used in their daily work. On the basis of their answers, the models were designed in an interactive way while being projected onto a big screen. Each modeling session took up to three hours, and at the end, all the process models that had been devised were shared among the participants (by email) for a final inspection and submission of

feedback. If necessary, the participants would agree to meet again at another modeling workshop to improve or change the model. With the introduction of the process-based modeling tool in QMP, each business unit now had to appoint someone to play the role of process owner. The responsibilities of this person were: (a) to control and (b) regularly update the process models of the corresponding business. The final versions of the models are then inspected by the process owners, who decide whether they are suitable and ready for the final ‘conformity check’. The employees of the quality management department then check the extent to which the process model complies with the requirements laid down by the aeronautical authority. Once this last check has been carried out, the process is integrated into the “repository” (process map) and “goes live” – i.e. it becomes publicly available and can thus be accessed by all the employees of AMC. In this way, during this procedure the figure of the process owner acts as a key component. As mentioned above, before QMP was started there was no conception of “business processes”, and the disconnected operation instructions and diagrams were largely ignored by the employees (whether consciously or not). With QMP and the modeling formalism employed, the assignment of an owner for each process – which was already inscribed into the modeling technique – is mandatory and binding. As one of the process owners whom we interviewed, stated: When there is ambiguity, for example, when one worker executes a process differently from what is specified, and I get to know about this, then I go to the worker and we discuss this. The process owner is thus responsible for the accuracy of the process model, which entails both ensuring that the model documents current work practices and seeing that everybody’s practice in fact complies with the model.

4.3 Examining the irreversibility of networks: the many facets of flexibility According to the quality management department, the QMP approach is based on flexibility and the number and nature of the model processes, organization charts, and roles are all constantly changing. The members of the quality department emphasized this dynamic character several times in the interviews, e.g. “the process world never stops!” Moreover, the factory floor workers we interviewed confirmed that changes are frequent in the process models of AMC.

When accommodating changes, the quality management software tool provides revision and feedback functions by means of which any employee can request updates to the model. The requests are first checked by the process owner, and if considered reasonable, are passed on to the quality management department for the conformity check discussed earlier. After undergoing a successful check, the modified model is transferred to the repository. In the strictly regulated environment of AMC, most work procedures are standardized, which means that incremental innovations based on small changes become very important. As a quality coordinator pointed out: We need licenses for maintenance and repairs and we have a license for making improvements […]. , These are not real innovations, but rather small changes, for example in the assembly line, and they enable us to be better than our competitors. In this way, the process models remain (at least partially) negotiable in the QMP network, and provide each of the process networks with a certain degree of flexibility. Indeed, the channels are kept open as a result of the negotiations of representatives in the modeling workshops, and the opportunity that the software tool provides for any employee to make model update requests. This allows changes and incremental innovations to occur in the work practices of the organization, and thus prevents the process networks from becoming completely irreversible. The roles acquired by both the employees and the process models in the QMP network, are of crucial importance to preserve this level of flexibility. The employees do not participate in the project network as mere ‘performers’ of rational processes that have been designed by outside experts to optimal standards, but rather are employed as active problemsolving agents and process designers themselves. Hence, in the process networks the models act as devices for guiding practice (and not laying down fixed rules that have to be strictly followed), and these can be updated and renegotiated. Another interesting aspect of the way in which changes can be incorporated in the process networks, was raised by a team leader we interviewed: Have you been to the other shop floor? They are doing everything “lean” now. It’s great how they have managed to become more productive; the figures are impressive. I want to do the same here too. […] I can make the whole shop floor “lean” and keep the processes [i.e. diagrams] the same! The interviewee thus stated that there are many aspects of the current work practices which are not

displayed in the process model. These features can be changed without having to alter the modeled elements in the diagrams (in fact, he also listed a series of possible changes). From a sociomaterial perspective, it can be seen that, in establishing the process network in the modeling workshops, some of the features of the work practices involved are selected and aligned with the graphical elements of the diagram – these consist of ‘what’ an employee is expected to do. All that was left out of the model becomes the question of ‘how’ these modeled tasks are carried out. Before the drawing of a diagram in a given modeling workshop, the ‘what’ and ‘how’ aspects of the corresponding organizational practices were not clearly discernible. In fact, this division into a ‘how’ and a ‘what’ is a contentious area, and we observed many situations in the modeling workshops in which there were different opinions about what should be documented in the models and what should be left out. In this manner, the distinction between the ‘what’ and ‘how’ aspects of the organizational practices, is in fact an important outcome of the business process modeling. And it is even more important if we consider the different degrees of flexibility that the ‘what’ and ‘how’ aspects entail. Apparently, it is easier to make changes in the non-modeled, ‘how’ features of the work practices. Since there is no prescription for them in the models, the process network can be correspondingly changed without affecting the arrangements inscribed in the model. However, after a careful examination of the updating procedure described above, a third factor related to flexibility is apparent. In the project network, the ways of introducing changes into process diagrams are not arbitrary, but must follow a clearly determined procedure. The procedure to request a change has to pass through the following: (a) the software tool, (b) the process owner, and (c) the quality management department. The first interesting point to note in this procedure is the degree of irreversibility that can be found in the assignment of the process owner. Unlike what occurs with the remaining contents of the process models (i.e. the other “what” aspects), no revision mechanism is provided for the assignment of the process-owner. Additionally, each change request restates the association of the owner with her/his process by making sure s/he is aware of and complying with the changes introduced into the model. Thus, while the process models are amenable to updates and provided with some flexibility, the range of responsibilities delegated by means of the owner assignment has a higher degree of irreversibility. In

other words, the resulting responsibility of the owner with regard to his/her own process models, is much less flexible. In this way, the process owner becomes accountable for the processes s/he owns. However, the process owner is not accountable for all the aspects of the processes s/he owns. Indeed, the project network has configured the scope of accountability in such a precise way, that each group of actors must be accountable for specific aspects of the modeled processes. This fact was illustrated in practical terms by the following anecdote which is based on a true story related by an employee of the quality management department: Can you imagine what goes on here when an aircraft crashes somewhere? Everybody panics and asks: have we maintained this aircraft? What went wrong? We work with aircrafts from all over the world, and before the fact-finding committee shows up, the first thing managers do in such a situation is to check if the corresponding processes [i.e. the process diagrams] are correct/compliant. If they are, we can say: we did nothing wrong, some [shop floor] worker must have caused the failure. But the opposite also holds true: if the processes are wrong, the worker can say: I only did what was written here, it’s not my fault. This anecdote illustrates the high degree of irreversibility of the project network with regard to accountability. In situations of crisis like this, many competing translations arise: Who was responsible, i.e. whom should the failure be imputed to? Here we can see how the sociomaterial arrangement of the project network reduces the space for disputes about the accountability of the actors by making a designated actor group answerable for a particular aspect of the modeled processes. The quality department is responsible for assuring that ‘what’ is modeled in the diagrams is compliant with official regulations. In contrast, the process owner is accountable for the accuracy of ‘how’ the modeled activities are performed in practice, i.e. s/he must ensure that the current work practices are in fact carried out in the same way (or in a ‘near-enough’ fashion) as what is shown in the process model. Thus, as illustrated in the anecdote above, this distribution of responsibilities is the basis for defining ‘imputability’ for the occurrence of failures. If the model had complied with the official rules and regulations, the quality department personnel could have breathed a sigh of relief, since it could be shown that the failure was not their fault. However, the contrary holds true too: if the process models turn out to be non-compliant with the rules, then the consequences will affect the quality management

department. As a result, the accountability of these two actor groups that is laid down by the project network has a low degree of flexibility, as it cannot be easily negotiated.

4.3 Interpreting flexibility: exploring multiple dimensions On the basis of our analysis, we propose an interpretation of the implications of process modeling for flexibility, which is able to resolve the apparent contradiction between modeling business processes and maintaining flexibility in the organization. This contradiction is also reflected in the different facets of flexibility that emerged from our analysis in the previous section, according to which different aspects of the project network have varying degrees of flexibility. The concept of multiple dimensions is put forward as a means of making sense of the different facets of flexibility identified in our case. Accordingly, the relationship between process modeling and flexibility must be analyzed by taking account of the different dimensions engendered by the associations that compose the sociomaterial network in which the process diagram is embedded. This multi-dimensional interpretation of the implications for flexibility is based on the sociomaterial perspective of ANT outlined in Section 2. According to this, processes can be regarded as complex actor-networks, and their degree of flexibility should be assessed in terms of the degree of irreversibility of these networks. Some of the associations of a network can prove to be more irreversible than others and, as such, prevent the network from incorporating new translations of a certain type. As a result, there is a reduced degree of flexibility in one dimension of the network. However, the same network may still be able to renegotiate some of its translations or enlist other (types of) actors, and thus retain flexibility in another dimension. The point becomes clearer in the analysis of our case study. In summary, we found that three different dimensions of flexibility emerged from the project network: (1) ‘what’ activities and relationships are shown in the process models (which are kept relatively flexible by the fact that it was possible to update the models); (2) ‘how’ the modeled activities are carried out in practice (this is left out of the models and thus has a even greater degree of flexibility); (3) ‘who’ should be accountable for certain organizational activities (this dimension has a considerably low degree of flexibility).

With regard to the ‘what’ dimension, i.e. the activities shown in the models, a reasonable degree of flexibility was achieved because a systematic revision of the models was possible. Decisive factors in bringing this about were the roles played by the models and employees in the actor-network. In these, formal artifacts (process models) are not imposed by an “outside” actor (e.g. a consultancy firm), but produced together with organizational actors. This means that the associations in the network between the model features and organizational actors (that translate a certain ordering of these actors) could be reversed, i.e. renegotiated by means of revisions. As for the ‘how’ dimension of the process networks, the process modeling tends to impose less restrictions on the flexibility of these associations, since they are not directly related to model features. Nevertheless, this does not mean that all the organizational features of the processes that are not related to elements in the diagrams are only loosely connected and their associations are not binding. The accountability that emerges from the project – in the ‘who’ dimension – has proved to be very robust. Indeed, the formal attribution of ‘process ownership’ and the procedure of compliance assurance reduce the margin for disputes about the imputability of the actors in the event of a technical failure, i.e. it reduces the degree of flexibility in negotiating the extent of accountability.

5. Discussion and Conclusion This paper has sought to illustrate the use and value of adopting a sociomaterial perspective, based on the Actor-Network Theory (ANT), and provide new insights into the different implications of documenting work practices in process models and the implications for flexibility that result from them. The study makes clear that there is a benefit in adopting a sociomaterial perspective that integrates human and non-human (material/technical) elements. Since business process models often serve as a basis for automation, which is operated by different types of computerized information systems (e.g. Enterprise Resource Planning, and Business Process Management Systems), they thus form the missing link between the ‘social world’ of organizational routines, and the ‘technical world’ of Information Technology artifacts. Being a hybrid, sociomaterial object par excellence, business process models are often studied either from a technical perspective (e.g. to define workflow methods and software tools), or a social perspective (e.g. in the study of organizational routines). However, this a priori fragmentation of process modeling into technical and social issues fails

to take full account of the sociomaterial complexity that arises from the constitutive intertwining and coevolution of the postulated social and material worlds. One practical consequence of the ANT perspective is that it enables a fine-grained analysis to be conducted of flexibility in process modeling, as illustrated in our case study. Although formalization and flexibility are often considered to be antithetical poles of a one-dimensional axis [17], the case study that is analyzed above, sets out a different picture of the relationship between the formalization of work practices in business process models and organizational flexibility. By employing the concept of irreversibility, it can be seen that the ‘translations’ carried out in the formalizing process models have formed some associations which are more irreversible than others. On the basis of our analysis, we suggest that the flexibility brought about by process modeling is multi-dimensional, i.e. it must be analyzed with regard to the different dimensions engendered by the associations that compose the sociomaterial network in which the process model is embedded. Each modeling project might thus engender different dimensions of flexibility, depending on the configuration of the networks that are established around the models. In this manner, the results of this analysis clearly show that the relation between formalization and flexibility is much more complex than the “zero-sum game” between formalization (into ‘dead’ models) and flexibility (of ‘live’ processes) that is often taken for granted as being applicable in a social analysis of process modeling, e.g. [14]. On the other hand, the symmetrical approach of ANT also goes beyond technical approaches that treat flexibility as an attribute that can be achieved by means of configurations or the employment of specific techniques in the artifact [4]. The distinction between the ‘what’ and ‘how’ dimensions of work practices is not entirely new. It echoes Suchman’s classical distinction between plans and situated actions (cited above) [19]. It is also related to two concepts used in the technical research on business process management: “process type” (i.e. ‘what’ should be executed) and “process instances” (i.e. the actual executions of a process type) [35]. What is new in our study is that it involves a sociomaterial perspective based on ANT that enables us to be more specific about the different implications of the ‘what’ and ‘how’ dimensions for organizational flexibility based on a detailed analysis of the sociomaterial networks involved. Furthermore, it allowed us to identify the ‘who’ as another

dimension with particular implications for flexibility in the case at hand. It is impossible (and not our wish) to claim that the dimensions that have been identified in the analyzed project exhaust the possibilities for every process modeling project. In fact, the ‘what’ and ‘how’ dimensions are quite general and can be said to be present in every project that relies on process modeling. This is because every process model can be seen as a 'how' of a particular 'what', and one can envisage an arbitrary disintegration of this type of relationship, depending on the degree of granularity that is required in describing a business process. For this reason, one important outcome of our analytical framework is that it enables a relationship to be established between the degree of flexibility of a certain dimension and the sociomaterial network established during the modeling project. Moreover, depending on the configuration of the sociomaterial networks involved, in other projects the ‘who’ dimension might only play a minor role, while other dimensions that have not been identified here, may be of much greater importance. This type of analysis is clearly a further step towards revealing the sociomateriality involved in process modeling, an area that has often been ignored and not consciously reflected on, in conventional technically-oriented approaches. Only a process modeling practice that is aware of the effects that emerge from the models within the associated social practices, will be capable of identifying and dealing with their implications with regard to flexibility, in an appropriate manner.

6. Acknowledgement João Porto de Albuquerque would like to express gratitude for the support granted by FAPESP (processes 2011/23274-3 and 2012/18675-1) and by the Alexander von Humboldt Foundation.

7. References [1] Weske, M., Business Process Management: Concepts, Languages, Architectures, Springer-Verlag, Berlin, 2007. [2] Smith, H., and Fingar, P., Business Process Management: The Third Wave, Megahn-Kiffer Press, Tampa, Florida, 2003. [3] Chang, J.F., Business Process Management Systems: Strategy and Implementation, Auerbach Pub, Boca Raton, FL, 2006. [4] Nurcan, S., "A Survey on the Flexibility Requirements Related to Business Processes and Modeling Artifacts", Proceedings of the 41st Annual Hawaii International

Conference on System Sciences (HICSS 2008), 2008, pp. 378-378. [5] Becker, J., Bergener, K., Mueller, O., and MuellerWienbergen, F., "Documentation of Flexible Business Processes-a Healthcare Case Study", AMCIS 2009 Proceedings, 2009, pp. 93. [6] Sarker, S., Sarker, S., and Sidorova, A., "Understanding Business Process Change Failure: An Actor-Network Perspective", Journal of Management Information Systems, 23(1), 2006, pp. 51-86. [7] Albuquerque, J.P., and Christ, M., "Revealing the Sociotechnical Complexity of Business Process Modeling – an Actor-Network Theory Approach", AMCIS 2012 Proceedings, 2012, Paper 5. [8] Callon, M., "Techno-Economic Networks and Irreversibility", in (Law, J., 'ed.' Sociology of Monsters? Essays on Power, Technology and Domination, Routledge, London, 1991, pp. 132-161. [9] Latour, B., Reassembling the Social: An Introduction to Actor-Network Theory, Oxford University Press, Oxford, 2005. [10] Lindsay, A., Downs, D., and Lunn, K., "Business Processes—Attempts to Find a Definition", Information and Software Technology, 45(15), 2003, pp. 1015-1019. [11] Becker, M.C., "Organizational Routines: A Review of the Literature", Ind Corp Change, 13(4), 2004, pp. 643-678. [12] Cohen, M.D., "Reading Dewey: Reflections on the Study of Routine", Organization Studies, 28(5), 2007, pp. 773-786. [13] Feldman, M.S., and Pentland, B.T., "Reconceptualizing Organizational Routines as a Source of Flexibility and Change", Administrative Science Quarterly, 48(1), 2003, pp. 94-118. [14] Pentland, B., and Feldman, M., "Designing Routines: On the Folly of Designing Artifacts, While Hoping for Patterns of Action", Information and Organization, 18(4), 2008, pp. 235-250. [15] Albuquerque, J.P., "Rethinking Formalization Processes in Computerized Systems: Analyzing the CoEvolution between Software and Organizational Practices", RECIIS - Electronic Journal of Communication, Information and Innovation Health, 3(2), 2009, pp. 7-15. [16] Berg, M., "Of Forms, Containers, and the Electronic Medical Record: Some Tools for a Sociology of the Formal", Science, Technology & Human Values, 22(4), 1997, pp. 403-433. [17] Cobb, C.G., Enterprise Process Mapping: Integrating Systems for Compliance and Business Excellence, American Society for Quality, Quality Press, Milwaukee, 2005. [18] Barad, K., "Posthumanist Performativity: Toward an Understanding of How Matter Comes to Matter", Signs, 28(3), 2003, pp. 801-831. [19] Suchman, L., Human-Machine Reconfigurations: Plans and Situated Actions 2nd Edition, Cambridge University Press, New York, NY, 2007.

[20] Woolgar, S., "Configuring the User: The Case of Usability Trials", in (Law, J., 'ed.' A Sociology of Monsters: Essays on Power, Technology and Domination), Routledge, London, 1991. [21] Leonardi, P., and Barley, S., "Materiality and Change: Challenges to Building Better Theory About Technology and Organizing", Information and Organization, 18(3), 2008, pp. 159-176. [22] Orlikowski, W.J., "Sociomaterial Practices: Exploring Technology at Work", Organization Studies, 28(9), 2007, pp. 1435-1448. [23] Orlikowski, W.J., and Scott, S.V., "Sociomateriality: Challenging the Separation of Technology, Work and Organization", The Academy of Management Annals, 2(1), 2008, pp. 433-474. [24] Østerlund, C.S., "Documents in Place: Demarcating Places for Collaboration in Healthcare Settings", Computer Supported Cooperative Work (CSCW), 17(2-3), 2007, pp. 195-225. [25] Lee, N., and Hassard, J., "Organization Unbound: Actor-Network Theory, Research Strategy and Institutional Flexibility", Oganization, 6(3), 1999, pp. 391-404. [26] Latour, B., Pandora's Hope: Essays on the Reality of Science Studies, Harvard University Press, Cambridge, Mass., 1999. [27] Law, J., "Notes on the Theory of the Actor-Network: Ordering, Strategy and Heterogeneity", Systems Practice, 5(4), 1992, pp. 379-393. [28] Law, J., "Actor Network Theory and Material Semiotics", in (Turner, B.S., 'ed.' The New Blackwell Companion to Social Theory), Wiley-Blackwell, West Sussex, UK, 2009, pp. 141-158. [29] Latour, B., "On Actor-Network Theory", Soziale Welt, 47(4), 1996, pp. 369–381. [30] Reckwitz, A., "Toward a Theory of Social Practices: A Development in Culturalist Theorizing", European Journal of Social Theory, 5(2), 2002, pp. 243-263. [31] Østerlund, C., and Carlile, P., "Relations in Practice: Sorting through Practice Theories on Knowledge Sharing in Complex Organizations", The Information Society, 21(2), 2005, pp. 91-107. [32] Mcmaster, T., and Wastell, D., "The Agency of Hybrids: Overcoming the Symmetrophobic Block", Scandinavian Journal of Information Systems, 17(1), 2005, pp. 175-182. [33] Walsham, G., Interpreting Information Systems in Organizations, Wiley, Chichester, UK, 1993. [34] Klein, H.K., and Myers, M.D., "A Set of Principles for Conducting and Evaluating Interpretive Field Studies in Information Systems", MIS Quarterly, 23(1), 1999, pp. 6794. [35] Schonenberg, H., Mans, R., Russell, N., Mulyar, N., and Van Der Aalst, W., "Towards a Taxonomy of Process Flexibility", Proceedings of CAiSE'08, 2008, pp. 81-84.