INFORMATION AND ORGANIZATION Information and Organization 16 (2006) 212–250 www.elsevier.com/locate/infoandorg What is IT? SAP, accounting, and visibility in a multinational organisation Paolo Quattrone a,* , Trevor Hopper b,c,d a b Saı̈d Business School, University of Oxford, Park End Street, Oxford OX1 1HP, UK Manchester Business School, University of Manchester, Oxford Road, Manchester M13 9PL, UK c Stockholm School of Economics, Sweden d Victoria University, Wellington, New Zealand Abstract Recent work on Information Systems tries to reconcile the apparent homogeneity of Information Technologies (IT) with the heterogeneity of their use by recognising that users can render IT systems flexible and malleable. This paper advances theorisation of this apparent paradox by reflecting on the nature of IT, i.e. its ontology. Observations of an ERP (SAP) implementation in a large USA multi-national cast within Actor-Network Theory and Science and Technology Studies approaches help illustrate how an object like IT can possess diversity and heterogeneity whilst being a homogeneous and operative technology. The paper argues that IT appears homogeneous for it attracts and generates heterogeneous uses. This paradox is labelled ‘heteromogeneous’. An IT system is theorised as an absence which establishes a presence by mobilising and attracting other actors and technologies, in this instance accounting, seeking visibility in organisations. IT emerges from multiple and continuous translations involving customisations of SAP. Thus the definition of IT is neither stable nor singular across time and space, which enables IT and SAP to travel across organisations. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Objects; IT; Actor-network theory; Science and technology studies; Time; Space; SAP; Enterprise resource planning system (ERP); Accounting; Visibility; Multinational organisation * Corresponding author. Tel.: +44 1865 278808; fax: +44 1865 278958. E-mail addresses: Paolo.Quattrone@sbs.ox.ac.uk, paolo.quattrone@said-buisness-school.ox.ac.uk (P. Quattrone). 1471-7727/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.infoandorg.2006.06.001 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 213 1. Introduction Orlikowski and Iacono argue that much Information Technology (IT) ‘‘research draws on commonplace and received notions of technology, resulting in conceptualizations of IT artefacts as relatively stable, discrete, independent, and fixed. [. . .] IT artefacts [that] tend to be taken-for-granted or are assumed to be unproblematic’’ (2001, p. 121). This paper addresses this lacuna by speculating on, ‘What is IT?’, ‘What are its features?’ and thence ‘How are its heterogeneous usages and users reconcilable with apparently homogeneous and stable IT systems?’ The proposed solutions are informed by observations of an Enterprise Resource Planning System (ERP) – SAP1 – implementation in a large American Multinational (MNO). These endorsed Boudreau and Robey’s comment that: ‘Although such technologies may be designed and implemented to restrict user discretion, especially when former systems are removed . . . users . . . consistently found ways to overcome intended restrictions’ (2005, p. 14). Yet, ‘physical properties of artefacts ensure that there are always boundary conditions on how we use them’ (Orlikowski, 2000, p. 409; see also Kallinikos, 2004). This conundrum is at the heart of problems in defining IT. If implementation and enactment commonly define ERP features, often unexpectedly, how do these features constraining action come into being? How can they exist prior to action if they are created by action? One solution is to argue that users interpret and enact IT differently. For example, Law and Singleton claim ‘‘objects look messy because people have different perspectives on them. [The object] simply means different things to people’’ (2005, p. 333; emphasis in original). This posits an epistemological solution followed by Information Systems (IS) research utilising Gidden’s Structuration Theory, which argues that IT is defined by complex interactions between IT artefacts and the meanings users attribute to them. That is, objects are products of structures and action, and the ensuing ‘‘interpretive flexibility’’ enables people to use IT differently (Orlikowski, 1992, 2001) and ‘‘simultaneously enact multiple structures’’ (Orlikowski, 2000, p. 411). This paper builds on such work, which is superior to realist approaches that assign pre-defined and fixed features to IT, to develop an alternative way of reconciling this multiplicity with the apparent homogeneity of IT. This theoretical exercise addresses several issues. The first concerns whether change is linear, i.e. do interactions between structure and agency traverse a linear timeline (Barley, 1986; Markus & Robey, 1988; Orlikowski, 1996; Robey & Boudreau, 1999), which attributes a clear and stable ontology to the IT undergoing change. The second issue concerns whether agency is granted only to users and not technology. Users’ interpretive flexibility enables IT artefacts to acquire different usages and features but IT features change only through this flexibility (Orlikowski, 1992; Orlikowski & Barley, 2001). The third concern is the theoretical possibility of distinguishing between IT artefacts (a structure) and users (the agents) (DeSanctis & Poole, 1994; Poole & DeSanctis, 2004). If IT systems are multifaceted due to potentially infinite user enactments (Orlikowski, 1992, 1996, 2000) then this distinction looses heuristic value. 1 SAP is an acronym for Systems, Applications and Product in data processing. The original German acronym stands for Systeme Anwendungen und Programme in der Dataenverarbeitung. 214 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 In summary, the ontological status of IT systems remains problematic. Hence, the advocacy of Actor-Network Theory (ANT) (Law and Hassard, 1999; Latour, 1987, 1999, 2005) and Science and Technology Studies (STS) (de Laet & Mol, 2000; Mol, 1999; Woolgar, 1988) to explain how IT materialises. IS research should shift from ‘‘multiple interpretations of objects [. . .] to thinking about multiple objects themselves’’ (Law & Singleton, 2005, p. 334) (see also de Laet & Mol, 2000; Dugdale, 1999; Latour, 2002; Mol, 1999). Here defining the nature and evolution of IT are seen as ontological rather than epistemological problems, partly because technologies, not merely people make objects (Latour, 1987). This requires more complex explanations of how IT becomes an ‘object’ (the ‘it’) than current IS research provides, for IT artefacts are less than a reified physical entity (they do not constrain behaviour entirely) but more than social constructions or individual enactments (they can prevent users doing whatever they wish). The arguments of this paper reside between this paradox’s intersects. The argument is pursued firstly by challenging some epistemological assumptions of Structuration Theory researchers within IS. They leave open important issues about their linear view of change, their treatment of agency, and analytical distinctions between IT as artefact and its users. A review of how ANT and STS research depicts the constitution of ‘objects’ lays a platform for ontological proposals for defining IT in this paper. Section 3 clarifies the research methodology, which does not claim to apply theory to case study observations but lets theory emerge from them. Section 4 outlines the resulting theorisation. Initially SAP was seen as a minimal actor that paradoxically became established by its relative absence. SAP proved difficult to understand as a working practice, which prompted translations2 involving customizations to deal with accounting and visibility issues. Each boundary making act to render IT a finite object implicitly re-opened it, making it impossible to treat technology as a black-box (Latour, 1987) or a ready-to-be-enacted artefact (Orlikowski, 1992). Section 5 returns to the central question, ‘What is IT?’ and delineates an alternative model. IT’s emergence is not attributed to institutionalisation that crystallises and homogenises IT’s components (à la Berger & Luckmann, 1966), which unwarrantedly labels some uses as deviant according to a prevailing structure. Nor is IT’s constitution as an object attributed to its homogeneity but to its heterogeneity, which attracts diversity. Paradoxically, this makes IT appear homogeneous, a phenomenon we label ‘heteromogeneous’.3 The conclusion summarises the main findings and delineates implications for future research. 2 Translation is a: ‘‘displacement, drift, invention, mediation, the creation of a link that did not exist before and that to a degree modifies the original [design].’’ (Latour, 1999, p. 179); see also Latour (1986, 1999) and Callon (1981, 1986). This concept is illustrated in relation to SAP later. 3 The structure of the paper and the proposed model (see Fig. 1) follows a linear path from ‘Absence’ to ‘IT’ due to the structure required by an academic paper. However the starting point and the sequence can differ as is clarified later. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 215 Fig. 1. The emergence of IT as a ‘heteromogeneous object’. 2. Theorizing the IT artefact 2.1. Change, structure and agency Orlikowski and Iacono (2001) observe that definitions of IT are often addressed in research on change.4 However, ‘What is change?’ is often taken-for-granted and selfevident in accounting and IT research (Quattrone & Hopper, 2001). For example, Barley (1986) portrays organisational change as the product of interactions between organisation structures and human agency. Here time (t) is linear – from t1 when the organisation had no IT system to t2 when it had. Epistemologically this presumes that both objects (IT and the organisation) had a clear morphology on both occasions. Quattrone and Hopper (2001, p. 408) argue that this is a form of ‘‘positivism for it implies that researchers or managers can identify these two separate states and the in-come and out-come of the ‘change’’’ process – i.e. they can view the entire process from a distant and detached space. Instead, like Ciborra (2000), they argue that change is non-linear for what is changing, who is observing the process, and their relation to features of the object undergoing transformation and its enactment at various locations at any one time may differ. The presumption that IT triggers organisational change, and IT is a managerial tool for executing its instigators’ intentions is deterministic and simplistic (Orlikowski, 1992, 2000; Robey & Boudreau, 1999). Several researchers searching for more complex interactions between IT and organisations have turned to Giddens’s Structuration Theory (1984) (e.g. Poole & DeSanctis, 2004). For example, Orlikowski’s technology-in-practice (2000) 4 Prime questions are: how to define objects undergoing change (IT); the effect of human agency and organisation structures; and whether IT is an emergent, dependent, or independent variable (see Markus & Robey, 1988). IT studies in this vein focus on outcomes (intended or unexpected) (Robey & Sahay, 1996), resistance (Robey & Boudreau, 1999), and whether IT is a driver, enabler or merely a tool for organisational change. They have opened up debates on ‘technological determinism vs. human agency’ (Orlikowski, 1996), and which theory best explains relations between IT and organisations (DeSanctis & Poole, 1994; Orlikowski & Robey, 1991, 2000; Poole & DeSanctis, 2004; Whitman & Woszczynski, 2004). 216 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 denies structures are ‘embedded’ in technologies but claims they are ‘enacted’. ‘‘Technologies are [. . .] not external or independent of human agency; they are not ‘out there’, embodied in technologies simply waiting to be appropriated. Rather they are virtual, emerging from people’s repeated and situated interaction with particular technologies’ (ibid, p. 407). Technology-in-practice helps explain how IT can deviate from designers’ intentions (Boudreau & Robey, 2005; Kallinikos, 2004; Schultze & Orlikowski, 2004) but whether structure and agency relationships can fully explain IT’s emergence as an object is debatable. Orlikowski illustrates how ‘‘people’s situated and recurrent use of a technology simultaneously enact multiple structures along with a technology-in-practice’’ (2000, p. 411). But if there are as many technologies-in-practice as people enacting the structure (as Fig. 2, ibid, suggests) then the ‘structure’ concept’s heuristic value is questionable (Giddens, 1989; Thompson, 1989). This also posits questions regarding agents, agency, action, and their interaction. Treating agents as unitary actors, individual or collective, fails to recognise they are ‘an association made up of elements which can be redistributed’ (Latour, 1991, p. 109). Moreover, some elements may be non-human (Jones, 1999; Walsham, 1997). Latour clarifies this when surmising whether: You can, with a straight face, [keep] . . . hitting a nail with and without a hammer, boiling water with and without a kettle, fetching provisions with or without a basket, walking in the street with and without your clothes, zapping a TV set with or without a command, slowing down a car with or without a speed-pump, keep truck of your inventory without a list, run your company with or without book-keeping . . . (2005, p. 60; emphasis in original). The obvious negative response does not imply that objects or subjects have agency or not but rather that agency extends beyond human responses to IT artefacts (Orlikowski, 2000, p. 425, note 3). The actor is instead an actor-network: agency and an object’s identity reside neither in an individual nor a technology but in a chain of relations between actors (and actants)5 (Latour, 2005). Discovering the nature and evolution of technologies requires disentangling this network rather than heuristic solutions based on notions of ‘structure’ and ‘agency’. This epistemological approach raises problems not solutions for unravelling IT’s ontology. 2.2. Analytical distinctions: the artefact and its use This raises issues concerning the dynamics between change and stabilization of IT, and an artefact and its use, and conceptualising these analytical distinctions, which reverts to the question, ‘what is IT?’ If its ontology needs definition but resides in constant enactment (Quattrone & Hopper, 2001) then dichotomies between stable and changing objects is unwise. 5 Latour defines ‘actant’ thus: ‘‘Instead of starting with entities that are already components of the world, science studies focuses on the complex and controversial nature of what it is for an actor to come into existence. The key is to define the actor by what it does – its performances – under laboratory trials. Later its competence is deducted and made part of an institution. Since in English the word ‘‘actor’’ is often limited to humans, the word ‘‘actant’’, borrowed by semiotics, is sometimes used to include nonhumans in the definition’’ (1999, p. 303). P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 Fig. 2. The ‘Green Sheet’: SAP, Accounting, and Spreadsheets. 217 218 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 IT research using Structuration Theory is sensitive to this temporal problem. For example, Orlikowski recognises that: ‘technologies are [. . .] never fully stabilized or ‘complete’, even though we may choose to treat them as fixed, black boxes for a period of time. By temporarily bracketing the dynamic nature of technology, we assign a ‘stabilized for now’ status (Shryrer, 1993)’ (Orlikowski, 2000, p. 411; see also, Orlikowski & Yates, 2002, p. 687). She continues: The practice lens elaborated here recognizes that even as technologies-in-practice may become institutionalised over time, this is only a stabilization for now. Every engagement with a technology is temporally and contextually provisional, and thus there is, in every use, always the possibility of a different structure being enacted (Orlikowski, 2000, p. 412). However, how can we assume a technology is stable (even for a nanosecond!) without reifying technology and limiting enactment possibilities? We wish to avoid the epistemological risk of having users and researchers examine a ‘thing’ – even if only momentarily, and analytical distinctions between the observer and observed. Separating technology from its continuous enactment delimits the latter’s role (Varela, Thompson, & Rosch, 1991). This permits Structuration Theory to accord time a central role, whereby interactions between structure and action follow a linear and recursive path in a progressive and ordered manner so that t2 always follows t16 (see DeSanctis & Poole, 1994; Orlikowski, 1996; Orlikowski & Yates, 2002). Case studies utilising Structuration Theory are often longitudinal to demonstrate how IT evolves over time (e.g., Burns & Scapens, 2000; Boudreau & Robey, 2005) but IT is not necessarily a single, stable entity at a point of time. For example, as illustrated later, SAP may not smoothly evolve over time (or space) but can acquire different forms simultaneously in different parts of the MNO. Moreover, SAP’s emergence as working practice relied on human enactments and non-human praxis involving cost calculations, legal reporting issues, and management controls (Quattrone & Hopper, 2001, 2005). These theoretical concerns prompted reflection on the ontology of IT informed by recent ANT/STS research on ‘objects’. 2.3. ‘Objects’ and theorising IT Law and Singleton in a special issue of Organization (2005) on ‘Objects’, argue that ‘‘social sciences methods are ill adapted for studying complex and messy objects’’ (2005, p. 331) and their review of ANT and STS explores fresh ways for theorising them. This section applies their arguments to the ontological question driving this paper. Law and Singleton encountered difficulties defining a strange object – alcoholic liver disease. This is not an object that IT scholars normally study as it is not an artefact and thus arguably not an object. However, Law and Singleton (2005) claim this disease 6 Orlikowski, 1996, p. 70, Fig. 2, uses arrows to depict interactions between organisational properties and organisational members’ actions. These cannot be reverted as structuration always leads to stabilization. More recently she affirms that the: ‘ongoing enactment of a technology-in-practice reinforces it, so that it becomes regularized and routinized [. . .]. Continued habitual use of a technology will tend to re-enact the same technologyin-practice, thus further reinforcing it over time so that it becomes taken-for-granted’ (Orlikowski, 2000, p. 410). Elsewhere we argue that if a structure, a technology, or an object is taken-for-granted then important facets of change and enactments are missed (Quattrone & Hopper, 2001). P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 219 is an object for an irrefutable reason: it kills. They initially tried to make sense of it by following the ‘epistemological strategy’, described previously, of using ‘boundary objects’ because the object proved complex and meant different things to different people. A boundary object is: Plastic enough to adapt to local needs and the constraints of the several parties employing them, yet robust enough to maintain a common identity across sites. [. . .] They have different meanings in different social worlds but their structure is common enough to more than one world to make them recognizable, a means of translation. The creation and maintenance of boundary objects is a key process in developing and maintaining coherence across intersecting social worlds. (Star & Griesemer, 1989, p. 393; emphasis added). Bechky (2003) provides a useful illustration. She found different professional communities referred to a machine differently. Engineers depicted it as a static entity by describing its components, whereas assemblers portrayed it dynamically referring to its stage in production processes. However, despite interpreting the machine (the object) differently and the difficulties encountered in reconciling this, they still referred to it as if talking of the same thing. IT systems can be understood similarly and IS researchers also have used ‘boundary objects’ to explain why IT can have multiple meanings whilst being sufficiently coherent to be discussed and acted upon (Star & Griesemer, 1989; Fujimura, 1992). Advocates of this approach argue this apparent contradiction enables IT and accounting practices to travel and engage various constituencies (Bechky, 2003; Briers & Chua, 2001; Carlile, 2002, 2004; Pawlowski & Robey, 2004). However, Law and Singleton (2005) propose a different theoretical strategy: these differences may be due to the nature of the object itself rather than multiple interpretations of it by users. This requires speculation on the ontology of objects. ANT has a long history of studying and theorising objects based on semiotics, i.e. viewing them as ‘an effect of relations with other entities’ (Law, 2000, p. 3). A word acquires meaning from its relation to other words (see Morin, 1977). Analogously, ANT recognises that objects are defined relationally: an object is never ‘out there’, i.e. standing proud of relations. Latour (1997), for example, notes when examining the journey of a TGV, the French bullet train, from Paris to Brussels that for the TGV be an ‘immutable mobile’, i.e. an object displaced without apparent deformation along the journey, the entire railway system must be well aligned, which requires considerable effort (Latour, 1997). If this network of relations is effectively aligned then the train is seen as a working technology – an object. Thus scientific and technical research is most successful when it becomes a ‘black box’, i.e. its content remains invisible. As Latour contends: When a machine [or an IT system] runs efficiently, when a matter of fact is settled, one need focus only on its inputs and outputs and not on its internal complexity. Thus, paradoxically, the more science and technology succeed, the more opaque and obscure they become (1999, p. 304). However, theorising this apparent immutability is difficult for the ‘‘unity of an innovation [technological or not] is not given by something which would remain constant over time, but by the moving translation of what we call [. . .] a quasi-object’’ (Latour, 1991, p. 117). Objects are thus never fully black-boxed. ANT theorists append the prefix ‘quasi-’ to overcome the ontological problem of defining objects that do not exist outside 220 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 networks of relations and thereby require something else to be defined. The prefix symbolises the problematic nature of black-boxes for they leak, thus they are never hermetically closed (Callon & Latour, 1981). de Laet and Mol (2000) accommodate these difficulties through the concept of a ‘mutable mobile’: An object or a class of objects [which] may be understood as a set of relations that gradually shifts and adapts itself rather than one that holds itself rigid. Thus while [ANT’s] intuition about the importance of relations was right [. . .] it got itself too concerned with standardization, with the rigidities of immutable mobiles that, if they exist at all, exist within rather specific networks that try to reach out over long distances and achieve centralised control (Law & Singleton, 2005, p. 339) The theoretical puzzle of defining objects from a post-structural, relational perspective is reconciling the homogeneity and heterogeneity of objects without granting too much agency to humans (as with boundary objects) or to objects (as in realism). As Law and Singleton (2005) state, ANT views objects as made in and by relations. However, how can the problem of having a ‘thing’ made by another ‘thing’ be avoided? Failure to resolve this multiplies the ontological problem this paper addresses. Objects should be more than social construction (to avoid relativism whereby anything goes and humans are king) and less than realism (to avoid reifying objects and ignoring interpretive flexibility). Theorisation should lie in between these two approaches. Latour (2001) addresses this using Tarde’s ‘reverse reductionism’. This reduces explanation to a single object, i.e. the ‘monad’, whose definition is complex ‘‘since the smallest entities are always richer in difference and complexity than their aggregates’’ (ibidem). The monad, like Latour’s conception of the object, is complex because the small always contains the big and it differs from itself as it is already other. In Tarde’s words: [The monad] is a milieu that is universal or that aspires to becomes such, a universe in itself, not only a microcosmos [. . .] but the cosmos conquered in its entirety and absorbed by a single being (Tarde, 1999, p. 57 quoted in Latour, 2001, p. 118). This definition replaces Euclidean views of objects existing in singular spaces with one where they are made in and by multiple spaces. Following Mol (1999, p. 77), an object is not differentiated by multiple features within a single reality but each feature is a different but inter-related object, which produces a multiple reality. In this milieu, linear time is unnecessary to explain how an object can differ from itself for it is already and potentially other than itself. Like a fractal (an image for which any suitably chosen part is similar in shape to a given larger or smaller part when magnified or reduced to the same size, English Merriam-Webster Dictionary On Line), the monad is an object made of infinite other objects which appear similar but make an object that differs from all of them. Understanding how IT evolves ‘over time’ and becomes different cannot only be attributed to humans’ ‘interpretive flexibility’ but also to its ontology, as IT is already many things. However, recognising that an IT object is already and potentially anything else does not resolve the ontological problem of combining multiplicity and stability. De facto an object does not become so just because it acquires presence whereas others fail by not doing so. For example, in our case SAP became the system used in the MNO but it left a niche for a different application with a similar function, SCALA, which was not fully integrated and accessible in real time. However, SCALA survived. SAP was clearly more ‘successful’ than P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 221 other IT systems used but this did not mean that it was homogeneous or stable. Thus, the theoretical problem is to understand how homogeneity and heterogeneity coincide in IT, and specifically SAP, as pursued in Section 4. However, prior to this, methodological implications of seeing objects as intrinsically other than themselves needs examination for, we argue, seeing time as linear does not help understand IT but rather is a problem to be overcome. 3. Research methods 3.1. The role of case study research in theorising: an epistemological position7 Methods are crucial for establishing a theory’s scientific credentials but they are not neutral (Latour, 1988): they are not the solution but should be the object of study when researching how scientific knowledge is constructed (Woolgar, 1988). Methods and science are black boxes that require unpacking (Latour, 1987, 1999). This paper is about objects but, following Orlikowski and Iacono (2001), IT artefacts, theories, and methods are not taken-for-granted, immutable black boxes for transporting them without deformation is impossible – ‘so rare, so miraculous’ (Latour, 1997, p. 183 see also 1999). For example, adopting a ‘theoretical framework’ or ‘grounded theory’ or ANT to make sense of our case would contravene the epistemological stance adopted. The aims here are more modest than grand aspirations of modernist approaches (Latour, 1991). If the observer and the observed, and the object of study and the methodology of its investigation cannot be divorced (Woolgar, 1988), then theory rests in its etymology. Theory is a way of seeing that blinds us to other vistas. Hence this paper is one view of how accounting and SAP operates, namely by offering visibility to business operations that incompletely represent organisational worlds (Quattrone & Hopper, 2005). This renders both IT and accounting technologies in flux whilst they remain recognised as objects. The paper demonstrates how reciprocal relations between accounting and SAP relate to visibility and action at a distance, i.e. the process of accumulating information in a centre of calculation to exert control (Latour, 1987; Quattrone & Hopper, 2005; Robson, 1991, 1992) and how IT is a ‘heteromogeneous’ object. The subsequent concentration on accounting and visibility lies in the researchers’ preoccupations and methodological predilections and how these became evident when studying SAP. Interest in management control in MNOs (Quattrone & Hopper, 2001, 2005) and action at a distance led us to engage with specific networks of actions (Czarniawska, 2004) concerning ERP technologies, accounting, and accountants. Other researchers driven by other issues would see a different organisation despite it being the same legal entity. MNOs (like any object – including IT) do not independently exist outside acts of engagement (Jones, Mclean, & Quattrone, 2004). A conventional section on data collection and analysis is not provided for data and observers are inseparable. As the Latin etymology of datum suggests (i.e. not only a ‘given’ but also a meaning given by the observer), they do not sit waiting collection (a practical impossibility given the vastness of MNOs) but derive from complex fabrication processes involving humans (e.g. the availability of researchers and interviewees, secretaries who arranged meetings and booked flights) and 7 Section 3.1 largely draws on Quattrone (2004a, 2006). 222 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 non-humans (e.g. theories, tape recorders, travel grants, note pads, means of transportation) (Latour, 1999). Scientific attempts to neutrally represent organisational realities are no different from other narratives despite their technical and rhetorical devices to pretend otherwise (McCloskey, 1985; Young, 2003; Callon, 1981; Latour, 1986; Czarniawska, 1998, 1999; Stone, 2001). It is more fruitful to explain the researchers’ pursuit of the case rather than assessing whether it faithfully represented happenstances in a SAP implementation. 3.2. Details on the case study and rationale for data selection The research studied a SAP implementation in a large American manufacturer and supplier of building products and composite materials internationally. Semi-structured interviews were held across the major business functions (see Appendix 1 for details). The interview protocol covered questions such as: ‘How does the evolution of information and cameral technologies affect management control in large organisations?’ Other questions were, ‘How to define the operation of management control systems in MNOs?’ and, ‘How does the organisational complexity of MNOs affect control?’. However, it was used primarily to start conversations. Interviewees were encouraged to speak freely: the emphasis lay on listening to how they mobilised abstract notions of management control and IT when enacting SAP rather than classifying responses according to preconceived factors. An epistemological premise about the case study duration is called for. It is commonly argued that longitudinal case studies help understand how the object under scrutiny evolves (in our case SAP) (see Burns & Scapens, 2000; Boudreau & Robey, 2005) but the epistemological implications of this are not always clear. A longitudinal case assumes a linearity of time and evolution which was not witnessed here. As reported elsewhere (Quattrone & Hopper, 2001), the complex intertwining of various IT systems operating in this MNO suggested that SAP assumed various evolutionary stages simultaneously and not necessarily progressively. The organisation was so vast that roll outs in different divisions and subsidiaries proceeded at different paces and did not necessarily follow the same pattern. As Mol noted, SAP was a complex object made by ‘‘different versions, different performances, different realities that co-exist in the present’’ (1999, p. 79). SAP was defined in a semiotic space (Law, 2000) containing continuous acts of engagements. Thus, the object did not need to ‘evolve’ over time to change but only required the enactment of different realities and connections with different practices (see Law & Singleton, 2005, p. 336) including, as demonstrated later, accounting. Change did not happen on a linear timeline but in a semantic network of relations that created multiple spaces and times (Quattrone & Hopper, 2001). This has implications for the rationale of longitudinal case studies. The passage of time was important for the researchers to gain an understanding of a strange object, notably SAP, but it did not help delineate SAP’s features for they did not exist outside relationships, including ones with the researchers. As Law and Singleton note, researchers should ‘‘think more carefully about the nature of the objects in the world – about what counts as an object’’ (Law & Singleton, 2005, p. 334; emphasis added). To understand what counts, what matters, what mobilises and engages is messy and time consuming. Here, but only here, time matters. Information on the project’s length and data collected may aid others to reflect on the interpretive journey the authors undertook and their conclusions but they P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 223 are not a rhetorical exercise to illuminate an illusory external reality explicating SAP’s evolution. Given this caveat, the research project lasted three years but the case took eleven months. The first phase commenced in the MNO’s UK subsidiary where six managers were interviewed for an hour and half on average. The second phase involved a visit to the European regional HQ in Belgium. Interviews were held with two financial controllers of the European business and a manager responsible for corporate IS (lasting two hours on average). The case culminated with a visit to the MNO’s HQ in the USA. Members of the accounting, budgeting and planning team were interviewed along with managers responsible for IS strategy and SAP implementation throughout the MNO. Fourteen managers were interviewed for one hour each. The company supplied copies of reports produced by the ERP system or reliant on ERP data. This information was compared with findings from interviews and helped enrich observations. Interviewees often recalled events from as far back as five years when SAP was first implemented, which may raise doubts about the accounts’ reliability. However, given the non-linear approach adopted this is not a concern for their initial difficulties in defining SAP were similar to when we interviewed them – they knew more but still did not know enough. Relating to SAP was always difficult. The final caveat lies in the paper’s use of quotations from interviews. These do not allow interviewees to speak for themselves (people, like facts, do not do so) nor do they represent things as they ‘really’ occurred (this wrongly demarcates objects from methods and investigators). This paper is on how framing devices produce visibility in large organisations. Writing a paper is similar – authors exercise power to select and arrange material from the field. The interview extracts enable theory to emerge from the narration rather than fitting events into an independent theoretical framework, which entails an untenable separation between theory, the case, and its authors. 4. Exploring IT: making sense of a SAP implementation A wide range of organisations have adopted ERPs (Scott and Wagner, 2003) for reasons ranging from Y2K bug fears to overcoming fragmented legacy systems (Robey, Ross, & Boudreau, 2002). However, adoption is often attributable to ERPs promising to integrate business functions, centralise business processes such as invoicing, quicken information flows, and render activities more visible and thus controllable centrally (Dechow & Mouritsen, 2005; Elmes, Strong, & Volkoff, 2005; Granlund & Malmi, 2002; Kallinikos, 2004; Quattrone & Hopper, 2005; Scapens & Jazayeri, 1999; Newell, Huang, Galliers, & Pan, 2003; EAR, 2003). Potentially, ERPs make information available in real time to anyone with access. This collapses notions of time and space – a selling point of SAP – an ERP market leader. In the MNO studied, SAP was presented as a ‘‘common, global, [and] simple’’ system8 that would link corporate HQ to subsidiaries across the world. In a large MNO, concepts such as real time integration, long-distance control and visibility are ideals which few can claim to be against (Hansen & Mouritsen, 1999, p. 451). As the corporate slogan suggests, SAP was purchased to enact a new vision of control to enable 8 The Director of Electronic and New Digital Technologies in the USA and the Director of Shared Services in the UK made similar comments. 224 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 anyone (with access) to get information on any part of the MNO, anywhere, anytime. It was argued that real-time, global visibility would collapse traditional notions of time and space, making integrated, central management easier (Dechow & Mouritsen, 2005; Quattrone & Hopper, 2005). This was compelling. However, ideals are by definition abstract, they do not become ‘real’ unless operationalised, enacted, and practiced. For instance, the ideal of ‘integration’ (but this could be ‘global’, ‘common’, ‘simple’ ‘real time control’, and, as we recount later, SAP itself) can assume different connotations in different multinationals. In this MNO it implied restructuring business processes and practices, whereas in another it entailed hierarchical control, continuation of redundant business functions, and, ultimately, avoiding job losses (Quattrone & Hopper, 2005). Nevertheless, though people may refer to the abstract word ‘integration’, how this becomes real (whether singular or multiple) depends on a complex network of relations. As Law and Singleton note, the object ‘‘subsists in, and participates in the enactment of, entirely different spatial logics or realities, and those spatial realities have complex relations with one another’’ (2005, p. 348). When people discuss these ideals they do so in their absence (Quattrone, 2006). This occurred here. SAP emerged from its absence as a working practice: users initially referred to SAP but as an abstract technology. Its lack of functionality helped it establish a minimal presence. Accounting and visibility issues then mobilised translations which heralded in customisation to establish SAP as a working practice. The continuous definition constituting IT as an object is detailed in Fig. 1. The next sections examine this constitutive process by examining each component of Fig. 1. However, it must be emphasised, as illustrated in the model, that whilst each component is presented sequentially for ease of exposition they are co-determined, do not invariably occur linearly or in disparate parts of the organisation simultaneously, or take similar forms. The analysis commences by scrutinising how SAP’s absence helped establish its presence. The next section examines its struggles to maintain a presence, not least when confronted by accounting practices, past, present and potential, creating visibility throughout the organisation. The following section traces how alliances with other actants, especially IT software, and translation processes involving customisation shaped SAP and accommodated actors’ often emergent needs. The final section describes how defining SAP was ongoing and disparate in time and space. SAP only became an object through enactment. The conclusion reflects on this process in its entirety noting how continuous translation and enactment rendered SAP’s features constantly unstable yet it managed to remain an object with meaning that constituted a working practice. This ability to maintain heterogeneous diversity whilst appearing to be a homogenous object is labelled ‘heteromogeneous’. It is argued that this ontological status enables SAP to establish and maintain presences within and across organisations. 4.1. Talking about IT in ITs absence: SAP ‘vanilla’ as an absent presence The ideal of ‘Common, Global and Simple’ proved fraught when implementing SAP. Managers felt uncomfortable with the unfamiliar information from the standard format. As a manager noted: It would have been wonderful if we could have taken SAP vanilla [SAP in its standard configuration] and implemented it. . . . SAP is a German company and I’m sure P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 225 that when they designed . . . SAP . . . it was probably along the lines of a German company (Information Systems Business Relationships, USA HQ). We do not know whether SAP is designed with an idealised German organisation in mind or whether the manager had been to Germany9 but he could talk about ‘a German company’ without exactly knowing what he is referring to. Managers initially referred to SAP similarly. They could discuss it without knowing what it constituted. SAP differs from business software solutions designed for each customer’s needs: it is not ready-made. To become operative SAP often requires considerable customization, ranging from simple reconfigurations to complete rewrites of sections of the software code – ABAP10 (Kallinikos, 2004; Quattrone & Hopper, 2001, 2005; Robey et al., 2002). Although SAP resembles a standard package, it is primarily a ‘meta-software’ with an architecture that connects its database to its constituent modules (Blain, Dodd, & Sandison, 1998; Hernández, 1997). To become operative it requires considerable data specification (i.e. what data will be stored and how) and business process definition (i.e. how business transactions and reports modify and draw upon this data), which are costly and carry failure risks (Robey et al., 2002). Managers discussed potential features and virtues of SAP but in their absence, for SAP was not yet a working technology with clear features and functionalities (though it may be in a different network). It could only be inferred and referred to as abstract ideals yet managers could refer to SAP, which established its presence by precipitating enactment of a simple three letter acronym – common, global, [and] simple’’ (Woolgar, 1981). This enabled it to forge relations with potential users and other information and cameral technologies (such as Excel and Accounting) that stimulated the introduction of other abstract concepts that underpinned more concrete actions. ‘‘Presence, in short, depends upon absence (just as absence depends upon presence)’’ (Law & Singleton, 2005, p. 342). Colós (1996, pp. 169–171, drawing on Laclau’s work, 1990, on representing objects), argues that users discover the unknown (be it an abstract ideal or SAP) by making abstract concepts present. These concepts are never clearly and univocally understood but are inserted into working relationships and are constantly represented or made present again, which enables users to talk about, refer to, and act upon them. Had managers known what SAP was and did, they would not have needed to discuss it (and IT!). They had to establish SAP as a ‘thing’, to make it ‘IT’, to make it real. Thus SAP initially did not exist as an object with clear features and functions but it established a presence through vague ideals lacking clear content but which people could refer to. To become an object SAP required enactment, the mobilisation of which is investigated below. 4.2. Making objects: how visibility and accounting mobilise action Making business transactions and operations visible is a key feature of ERPs (Elmes et al., 2005) but SAP initially failed to do so in this MNO. To paraphrase Latour (1999, p. 24) ‘‘How SAP packed the world into numbers?’’ required unravelling how 9 If it did, this would be an interesting example of how a reference to something absent, the abstract ideal ‘German organisation’, would make designers ‘do things’ – in this instance, design the software’s architecture. 10 Acronym for Advanced Business Application Programming (Hernández, 1997, p. 23). 226 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 managers and organisational practices (especially accounting and IT) made abstract references to constitute objects. Other objects and actants had to be attracted and support mobilised to make practices visible. 4.2.1. What do we need to see ‘things’? A network How things are made visible needs addressing. Latour (1999, p. 24) denies any ontological gap between nature and science for science refers to objects but also constitutes them: ‘‘there is neither correspondence, nor gaps, nor even two distinct ontological domains’’ (1999, p. 24). He illustrated this with an account of a scientific expedition into the Amazon forest to establish whether boundaries between the forest and the savannah were advancing. He traces the shift from a photograph taken at the beginning, when the boundary was difficult to ascertain, to a written scientific report at the end making it definite. He examined how the boundary between the forest and the savannah was constituted by referencing (a fuller account, with pictures and photographs, is in Latour (1999, pp. 24–79)). An instrument called a ‘pedocompactor’ (a square wooden box with white paper squares to deposit soil samples) was vital. Once samples were collected using sampling theory developed by science they were put in the pedocompactor and became transportable. Back in the laboratory each sample could be coded according to a colour chart (The Munsell Code) and assigned a number. The numbered samples could then be written on a piece of paper. Once this chain of references was established the blurred distinction between two badly defined entities (the forest and the savannah) became a clear difference, i.e. a boundary on a sheet of paper. The references did not just refer to objects – they constituted them. Neither the forest nor the savannah was as visible in nature as on the map drawn by scientists using references. Inscriptions referred to the savannah and the forest in their absence. They became real only in the scientific report prepared at the journey end through a chain of transformations that constructed relations between absences. These made other abstract tools and practices present and gain sense when establishing the boundary between savannah and forest. The pedocompactor, the Munsell code, and even the scientists are present due to the absence of the two objects they were studying. Transformation chains, however, can collapse if the unexpected occurs, e.g. samples can fall to the ground, and ‘the world can return to confusion at any point’ (Latour, 1999, p. 39). Or it may prove impossible to reconstruct the chain of events necessary to establish borders between forest and savannah in the scientific report (Latour, 1988; Woolgar, 1988, 1991). We can examine SAP similarly. It encountered problems rendering things visible when first implemented: managers felt uncomfortable with the unfamiliar information provided. Even when its information was recognisable, its real-time integration and configuration of distances between HQ and subsidiaries violated how managers had exerted control using accounting numbers (references) to make activities visible (Quattrone & Hopper, 2001). Accounting, visibility, and SAP were intertwined and the continuous, reflexive emergence of new information requirements rendered SAP’s characteristics precarious and unstable. The UK Financial Controller recollected the ensuing chaos: Because [SAP] is a global system and because there is a lot of inter-company trading . . . as well as people in the UK posting in the books. . . people in Belgium, from Italy, from Spain, from anywhere . . . as long as they have got access rights, can affect my books. The big problem [is] that . . . you are reporting on a deadline and you think by P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 227 the end of the day when you have finished all your postings. . . . You know what result you have and you tell the States, ‘‘This is what we have got’’ and you will send them a flash report that says revenue x. They can come back to you five hours later and say, ‘‘We have just run the report and it is not the same report!’’ And then it is like, ‘‘Shit! Somebody has just posted something in the books!’’ . . . That process is being better managed now, but two years ago when I arrived I couldn’t believe it . . . it was like anarchy. People were posting into your books and you didn’t know about it. As he recounted, this created problems: I came from a business where everything was very tightly controlled. . . . Suddenly . . . everybody is impacting on your results and you are not quite sure why. It can be a real pain and a . . . huge embarrassment. We had one particular situation in Belgium where I reported a result . . . and somebody else in Belgium changed it by $300 000. I didn’t know that he had changed it; we spent two days trying to hunt this change down. Sounds like, ‘‘Why can’t you find $300 000 of change?’’ Believe me, it was bloody difficult. Apparently simple things like designing a report could provoke different views of what is ‘good’ or ‘bad’, as a UK manager pointed out: In the States they are very much keyed into . . . full absorption costing – trying to tie back to actual costs and assigning actual costs to each production order and each product to actual unit cost. Over here in the UK we are much more focused on standard costing . . . that was the big difference coming over here (Plant Analyst, UK subsidiary). Even determining transfer ‘prices’ (another absent presence) was problematic: When I talk about pricing I am talking about the price we get in the market place and when [the HQ] talks about pricing they talk about somebody putting a number into the system (UK Customer Service Leader, referring to internal transfer pricing issues). What was obvious and important to the UK subsidiary could be irrelevant at HQ. For instance: We would need a report that can extract the appropriate data, summarise it, and give it in manageable and recognisable format. . . . They were keen to avoid a situation where every part of the world went out and developed its own reports. What they were trying to do was get a global report. . . . That is useful in theory but not quite so useful in practice. . . . For instance, in the Continental business until the advent of the Euro the whole question of currency management was key . . . because . . . they were selling in seven currencies. They were talking to people in the States and asking: ‘‘The currency?’’ [They replied] ‘‘Well, it’s the dollar isn’t it?’’ (UK Financial Controller) SAP was received unfavourably because it did not use well-understood accounting inscriptions to represent activities previously deemed important. Users felt blind for they could not ‘see’ events in remote subsidiaries. Even tracing $300 000 back was ‘‘bloody 228 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 difficult’’ as the profanities of the Controller testify. This typified the problems many managers’ encountered when initially using SAP. Their response, like Latour’s scientists in the Amazon, was to try and order events by tracing them back to their origins. In Latour’s example sampling procedures were perceived as reliable. In contrast SAP generated accounting references that managers constantly questioned and distrusted, partly because to become operative SAP required considerable data specification (i.e. what data will be stored and how) and business process definition (i.e. how business transactions modify and draw upon this data). The initial installation of the standard format of SAP was technically successful but its output denied managers the comfort of knowing remote locations afforded by previous systems. It became apparent that the standard package would not tidy things up – instead accounting muddle reigned. Managers sought order and real-time visibility but instructions on how to achieve this using SAP did not exist. SAP became unexplored territory – the map needed to orient managers in the MNO’s ocean of transactions was as yet unwritten. In the resulting mapping accounting became an important player (albeit not the only one) for it can, ‘‘define roles, decisions rights, objectives and performance criteria (Mouritsen, 2005, p. 12). Accounting abstractions became vital for defining what should be framed and seen on SAP’s computer screen (e.g. a budget variance). Complaints about SAP led to a decision to involve managers more in determining what information it should provide. For SAP to become an operative IT system (a fact, a ‘thing’) giving visibility to other ‘things’ (e.g. costs, delivery times), it had to enact the acronym attributed to it and its associated ideals to attract and engage ‘other’ actors and actants. SAP had to engage with heterogeneous diversity to be perceived as homogeneous. If SAP was clearly defined people would either like it or not and the mess would be clarified. Its implementation would either encounter no problems (if users liked SAP – it would run as smoothly as a TGV running from Paris to Brussels, Latour, 1997), or it would not take place (as people would not like it). However, SAP being opaque was simultaneously absent and present – as denoted in the heterogeneous objects model in Fig. 1 – in order to generate its mobilization and action (Law & Singleton, 2005, p. 344). Even the reason for implementing SAP referred to something absent but present in its effects. The Director of Accounting at US HQ when asked why the MNO decided to implement an ERP system stated: ‘‘The main justification was we were not Y2K compliant’’. We now know that this bug was not a bug, and that on January 1st 2000 things continued to work regardless of safety measures taken by companies. Here an absence (the bug and its consequences) became present due to people referring to it. Talk about something and it will become real. 4.2.2. Accounting abstraction and its power to mobilize According to Laet and Mol a successful technology ‘‘must be capable of gathering people together and of inducing them to follow well-drafted instructions [. . .] it must seduce people into taking care of it’’ (2000, p. 235). However, SAP initially failed to grant users ‘hope’ that it would solve their problems (a key organisational feature, Jones et al., 2004) as it failed to establish an effective network of relationships between humans and nonhumans (such as accounting practices) and thus engage users. To overcome this SAP had to accommodate managers’ beliefs on what accounting should represent and thereby make visible. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 229 What is visible – what ‘images’ enter peoples’ consciousness – creates possibilities for IT to be mobilised, enacted and defined. Images (and imagination) are vital facets of interfaces between infrastructures such as IT and individuals seeking order (Mackenzie, 2003, p. 366). Such encounters require material practices (Dugdale, 1999), including accounting, to provide control at a distance by making things visible on a sheet of paper (Robson, 1991, 1992). Making items visible also involves absences, for one way of seeing precludes seeing something else. For example, in Escher’s lithography ‘Print Gallery’ (see Hofstadter, 1979) every representation involves a hole that can be made infinitesimally small but never eliminated. It can be moved across the picture to make some obscure parts clearer but simultaneously this obscures or modifies other parts. This hole, this absence, creates visibility. Similarly, we would not need IT (or accounting) representations to make things visible if they were already so. As the Director of Finance at the USA HQ commented this contains advantages and disadvantages: The bad of centralisation is that I have now . . . accountants most of which have never worked in a manufacturing facility. [On the other hand] I have got [a guy] who lived 20 years in a manufacturing facility and then came to a centralised function [in HQ]. . . .. When he is talking about a piece of equipment in the factory he knows what he is talking about. I’ve got other people who have spent a week at the most and some never more than a day at a facility. It is really tough to be a decent accountant, I think, without having worked in a manufacturing facility. His definition of a ‘decent accountant’ identifies an important accounting feature: charts of accounts, accounting theories and principles fabricate order from a mess of transactions through abstraction and synthesis (Armstrong, 2002). Accountants and SAP designers may rarely physically see the products or locations or people they represent.11 Their tools make abstract references to objects using entries originated by transactions denoting the existence of say products ‘in stock’ (another abstraction) in the guise of a number. Accounting describes costs, revenues, and so forth in their absence even when references like ‘inventory’ refer to concrete objects in the warehouse (see Macintosh, 2002 for issues about signs referring to hyperreality). IT and accounting are crucial in this absence/presence pattern for ‘‘while aiming to introduce order into certain areas of practice in information management, the data sets would simultaneously introduce disorder elsewhere’’ (Bloomfield & Vurdubakis, 1997, p. 655). In our case, accounting abstractions provided continuous ordering, i.e. they created references that people or technologies could refer to, in the absence of the object that accounting referred to (the referent). Accounting, like SAP, was a pattern of absent presences (Law & Singleton, 2005) and they were interdependent. For example, SAP relied on accounting inscriptions to make transactions visible, define their ‘poles’ (e.g. between buyer and seller, or controller and controlled), and provide techniques to connect them. In other words, SAP and accounting contributed to boundary making. For example, budgeting created distinctions between headquarters and subsidiaries, divisions and regions. Accounting contributed to complex processes of definition and mediation materialising millions of transactions into financial and management reports on SAP’s screens. 11 ‘‘Who has ever seen a cost?!’’ Anthony Hopwood, ad vocem. 230 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 The UK Financial Director explained how accounting indicators in what was colloquially known as the ‘Red Book’ gave guidelines on what SAP’s customization should make visible to SAP’s implementers. He stated: [The Red Book] tries to map . . . the usual things, sales profitability and return on capital but also things like price movements . . . how the sales price trend is going, what the cost price trend is doing, what our measure of what we call cost profitability is. The Director of Accounting in the USA HQ provided another example: I have got a list of two thousand, three thousand accounts and we have a definition that is known to these folks. It’s just now being put on the Internet, so they have access to that. Before we had a book. . . . Each account has got a definition, so account 50010, for example, is customer sales, in the US, in Brussels, in Hong Kong. Accounting categories extended beyond external financial reports to management accounting issues. The UK subsidiary’s Financial Controller remarked how: We agree what the core measurement tools are. . . . On the sales side we are looking at average sales prices in key sectors of the market. On the manufacturing side we are looking at the . . . summary of . . . variable costs per ton, . . . our production overhead spend, and then . . . our sales and general administration spend . . . Then you start to break that down into lower levels of detail. . . . Some of those key indicators are done on a monthly basis but some . . . are almost daily. For instance, the plant people [can] tell from week to week how their cost per ton is performing because they are getting that information very readily. Accounting does not mirror economic, social, and organisational realities – it is a framing activity (Bloomfield & Vurdubakis, 1997; Hines, 1988; Suzuki, 2003). Accounting categories (e.g. costs, revenues, and income), principles (e.g. feedback control), and techniques (e.g. double entry bookkeeping, cost allocation) transform desires for order (‘hope’) into actuality. For instance a budget is not a technical tool which is mechanically applied. Rather it provides users with an order that demarcates the present and future, and the controller and controlled. However, budgets are also means of political compromise and organisational negotiation (Hopwood, 1980). Hence accounting is not a static structure but a continuous process (Quattrone, 2004b). Once items are abstracted in accounting reports they can be referred to (whether a product is in stock or a worker on the shop floor) and people can use the reference, e.g. the accounting sign in the accounting report, in the absence of the object, i.e. the referent. This occurred in this MNO. Accounting did not immediately provide the visibility sought but prompted arguments for change involving customisation. Accounting interfered (Law, 2000) despite its principles having no predetermined essence (Miller & Napier, 1993; Quattrone, 2004b). A ‘cost’ is never an autonomous entity. Its meaning depends on a process of definition, as inferred in the aforementioned disputes about which transfer pricing or cost calculation (standard, full absorption, or ABC) systems should prevail. Processes of definition cover how a cost is calculated, whether it includes certain items, and the purpose of its calculation. Thus accounting simultaneously mobilised, i.e. ‘‘intervenes and creates supplements that change the conditions for the design to work’’ (Mouritsen, 2005, p. 14), and interfered by framing what was desirable, apparently achievable, and P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 231 counted as an object (Mouritsen, 2005, p. 13). However, this journey was long and required substantial customization of SAP, translations, and engaging other actors and actants, as is recounted below. 4.3. Customising SAP: Translation and engagement One SAP R/3 manual states: ‘‘Implementing R/3 basically means using SAP R/3 application software to solve the information needs of the business’’ (Hernández, 1997, p. 819). However, managers and software designers found this difficult as no investigation of managers’ information had been undertaken. When it ensued managers questioned established ways of conducting business. Thus mapping and incorporating user needs followed a nonlinear path and proved more difficult than anticipated. Managers discovered that determining ‘good’ information did not mean merely applying the standard SAP package or, for example, reproducing previous accounting practices within SAP, or even learning new optimal representations of the business. Rather it was a chaotic process of ‘drift’, discovery, and mediation (Quattrone & Hopper, 2001). People perceived SAP as less effective because its standard version provided unaccustomed accounting data. The Director of Accounting in the USA HQ commented: Well, I put in SAP and I had this great, integrated package. . . . Well, my costing system in SAP is not nearly as sophisticated as what I had . . . before. So I got sub-optimised as a costing person. . . . My accountants . . . were using ‘Focus’ . . . I hated Focus and the ledger system was bad but these guys loved it because they knew it. Laet and Mol in a study of Zimbabwe Bush Pump ‘B’ technology noted that it was ‘‘expected to perform. It must act, do something. It is made to work’’ (2000, p. 238). This is true for SAP and other IT systems: to be recognised as technologies they must function (to a degree). However, SAP failed to do so initially – thus its ability to engage, i.e. establish connections between the technology, users and other management practices like accounting, was limited. Thus it failed to attract others and gain credibility until customisation, via major rewrites of the ABAP code, became the norm not the exception. ABAP programmers, nicknamed ‘‘Abappers’’, became the new cartographers as only they could rewrite the software. The USA Director of Accounting described the customisation thus: There is taking SAP as it comes in and then kind of configuring it to say: ‘‘OK, I want this account, I want to be able to do this with my customers’’; ‘‘I want to have different businesses on one order’’, ‘‘I want to have this kind of shipment’’ – pretty basic things to change. The customisation at the other end though, [entails] writing the code that SAP is written in . . . called ABAP, which is a very difficult, I think, language to understand computer-wise. He continued: At the beginning we said, ‘‘We are not going to do that! [i. e. change the ABAP code]’’. But pretty quickly in the businesses they said, ‘‘Well that is not acceptable. I have got to have this for my customers’’. You then get sucked into making customisations really quickly. It was really tough for the IS folks and the project folks to say, ‘‘I am not going to do that.’’ I mean, if [the UK Controller] says, ‘‘Well, what we 232 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 need here is . . .’’ it’s pretty tough to argue with that from the States. . . . It’s pretty easy for a guy in Brussels, for instance, to say, ‘‘Well my local statutory needs are these and local government won’t allow me to do anything else’’. Belgian guys are really good at that. The ABAP code was important for SAP’s definition. Normally modifications are not common, for they require expensive professional expertise from ‘Abappers’ but they are not impossible. Elsewhere, Robey et al. observed: ‘‘the commitment to avoid reengineering was so strong that managers modified the ERP package in order to make it conform closely to existing organisational processes’’ (2002, p. 35).12 This happened in the MNO studied here. Enacting the idea of a ‘‘common, global [and] simple’’ system throughout the corporation was soon perceived as unachievable and economically infeasible. For example, SAP could not be established in an Italian subsidiary with employees who were not fluent in English (the language of SAP in this MNO). It was too expensive to adapt SAP into Italian and impossible for Italian employees to learn English quickly. Employees began to realise that the slogan ‘‘common, global [and] simple’’ had no substance outside its enactment (Varela et al., 1991). Customisation prompted a process of translation, i.e. a ‘‘displacement, drift, invention, mediation, the creation of a link that did not exist before and that to a degree modifies the original [design]’’ (Latour, 1999, p. 179). Physical modifications of the ABAP code reinterpreted the originally bold ideals of SAP into more familiar forms of visibility, which helped reconstruct shared understandings of users’ needs. SAP in its standard format and simple reconfiguration had been absent as an IT system but once managers realised it could be house-trained to reproduce locally acceptable controls they perceived it as less of a ‘‘monster’’ imposing homogenised business processes from HQ. Following customisation local managers realised how malleable SAP was and that it could frame order out of chaos (Bloomfield & Vurdubakis, 1997). Customisation brought reflection about the accounting information SAP should abstract from daily transactions. Ad hoc teams and centres were created throughout the MNO to enable personnel to redefine SAP. SAP began to engage differences through constant encounters with potential users and other technologies in the space afforded by these centres. Mediation and translation endlessly modified SAP as a ‘package’ but increased its presence. For example: Customer Pride [one of these centres] was set up. . . . They listen to what we have to say and what our priorities are. . . . So if we come up with a problem . . . that Customer Pride can’t solve or is new to us and to Customer Pride we can use . . . their people to work on it for us. (UK Customer Service Leader) Another centre reconciled employees’ different and shifting information requirements and associated accounting problems. If accountants wanted SAP to reconfigure transactions or change a report’s purpose, they could request an investigation by this centre at HQ. A plant analyst in the UK subsidiary recounted how: ‘‘There is a ‘wish list’ of reports that need to be written’’. The centre then customised the software to produce meaningful 12 Large MNOs have the resources and power to convince vendors that the system needs adjusting to the corporation’s business processes but a different situation prevails in smaller firms where resources (financial and expertise) are limited. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 233 accounting reports, often after mediation between local needs of the subsidiary and the global aims of HQ. The ensuing re-translation of the accounting requirements of both parties rendered SAP the object of translation. For example, getting managers to agree on a common template was difficult, as an IS specialist explained: Well we brought those people together and said, ‘‘You folks have got to agree on common processes before we can put this system in’’. For the most part we were successful . . . but it was sometimes a very emotional conversation. But you can’t use a common system in five different ways. You have got to configure the common system one way and adapt your local practices to match it. Initially many actors were uncertain what they wanted and what SAP could provide – (‘‘a lot of things’’, an interviewee mentioned). Mediation accompanying customization helped managers ascertain their information needs and the attendant confusion provided a platform for debate. Thus SAP began to engage what was different from it: not only with respect to potential users but also other technologies. SAP is an ‘open technology’ that uses ‘‘standard formats for data exchange, communication interfaces, and programme-to-programme communication’’ (Hernández, 1997, p. 20). When SAP failed to satisfy accountants and production managers locally it proved amenable to forging alliances with other software packages – in this instance Excel and house-made bolt-ons. SAP’s ‘open technology’ allowed such data exchanges, which helped reconcile emerging local needs with standardised global information requirements from HQ. For example, a UK subsidiary’s plant analyst downloaded data from SAP into Excel to develop his own control system (known as the Green Sheet – illustrated in Fig. 2), which became widely used within manufacturing plants of the Roofing and Insulation business. Its creator described it thus: The Green Sheet is exactly the key indicators that, from a manufacturing point of view, someone like [the financial controller] or [the manufacturing leader] would want to see, ‘‘How has [the plant] done?’’ They won’t run a P& L; they won’t look at a balance sheet [but] they will look at the Green Sheet because it is a snapshot view. . . . Every day I will be looking at each cost account and checking . . . that the costs. . .[of my plant] . . . have gone to the correct accounts and make adjustments as needed. Elsewhere employees used spreadsheets to make sense of data from SAP as the UK Financial Controller illustrated: Typically . . . an analyst . . . will run a very detailed transaction report within SAP, which may take 45 min and involve 2000 transactions. Then information will be dumped from SAP into something like Excel and it will be played around with there, reformatted, summarised, and . . . used as the management report. SAP doesn’t really give us management reports. The only exception . . . is an add-on to SAP, which is called the ‘‘cost and profitability analysis system’’. [It] does actually allow us to print fairly detailed summarised reports of sales margin performance either by market sector or product type or any which way you want quite frankly. But . . . if I wanted to get a purchase price variance report from SAP I can’t do it or not without a lot of manual interface which is running a transaction dump and playing around with it in Excel. 234 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 The costing system within SAP, Activity Based Costing (ABC), 13 also precipitated alliances with spreadsheets and prompted mediation that defined both SAP and ABC. A Plant Analyst in the UK subsidiary commented: I have to understand the cost drivers and know what the line speed is going to be running at, and therefore know the activity rates that are going to be in SAP, and how the cost is going to be accumulating. . . . I have to understand what SAP is costing, and . . . the manufacturing system, and match those two up so that we have a true reflection of the cost. The mediation that established ABC as a working practice prompted adaptations of both ABC and SAP to extend accounting beyond scorekeeping to facilitating problemsolving by managers who could reshuffle data within SAP (or by spreadsheets). In summary, SAP’s establishment required a ‘‘community’’ (de Laet & Mol, 2000, p. 245) that defines what is to be seen, how and why. This extended beyond users making abstract references to costs, ideals, and technologies to information and cameral technologies such as Excel and accounting, as users could not conceive or extract cost data without them. These required further mediation that co-defined users, their needs, SAP, and accounting. If SAP had been inflexible and did not engage this community – these ‘Others’ – it would not have established itself as an IT system (see Kallinikos, 2004), for examples of ERP implementation failures. The SAP implementation story could be depicted as a longitudinal process of changing the standard version of SAP through customisation and user involvement to a new version. However, SAP never originally existed as a tangible object with clearly defined features as explained previously. This argument is extended below. SAP never reached a single endpoint where it acquired stable and consistent features: it never became defined as a homogenous object. 4.4. The continuous tension of ‘de-finitions’: ‘SAP doesn’t stop!’ ‘To define’ comes from the Latin definire, which consists of the prefix ‘de-’ (‘to be about’ but also ‘to do the opposite’ – ‘to deprive’) and the verb finire (to limit – from finis – a boundary or end). This etymology illustrates the rich and conflicting nature of any definition: on the one hand, ‘to define’ establishes boundaries but on the other it demolishes them (Merriam-Webster English Dictionary). Every de-finition (a closure) is also a de-finition (an incomplete order). To become a working technology, SAP required many definitions – whether to configure SAP ‘‘vanilla’’ or to customise it via ABAP rewriting. However, this was not always evident as a manager’s remark illustrates: [Our company] conducts business transactionally, just like any other business in the whole world. Just like I do my chequebook at home. Just like . . . when you start peeling the onion and . . . that is all motherhood and apple pie. That’s the way it works (Director of Electronic and New Digital Technologies, USA HQ). 13 Their observations are pertinent for ABC was the accounting system in the Controlling module of SAP R/3 adopted by this MNO. However, our research did not examine ABC implementation and usage in detail. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 235 According to him SAP is a transaction processor and nothing is apparently simpler. SAP is assumed to be a ‘software package’ with established functions and features. How transactions require definition using accounting to constitute exchanges between HQ and organisational segments is taken-for-granted. However, this was not inevitable. For example, buying and selling, parties to transactions (the entities involved), transactions and entities rendered by legal or man-made boundaries (often through accounting), their connection via transfer prices (market or negotiated), and agreements defining transactions (instalments, exchanges of goods, leasing) can vary, with repercussions for accountability and what becomes visible, manageable, and controllable. For example, a spare part purchase may be classified as maintenance and repairs and charged against a production department’s operating costs, or it can be attributed to a design error, charged to an engineering department, and classified as a capital cost. How this transaction is represented influences performance measures and accountability within the firm. Moreover, who can post, classify, access information, and when, has management control consequences. Computer technologies such as ERPs may be replacing hand-written ledgers but seeing things through double-entry bookkeeping remains an issue (Carmona, Gutiérrez, & Cámara, 1999). What constitutes ‘good’ and ‘bad’ transacting can differ, as the Director of Finance ruminated: When I worked for [another MNO] and . . . became involved in the Middle East. . . . There were business practices. . . I was most uncomfortable with . . . [but] that is just the way they do business. The quotation may be a stereotype but it illustrates how apparently simple activities, like transactions, assume meanings often ignored by system designers. Moreover, defining a transaction during ERP design and implementation is a political not a neutral act with repercussions throughout the organisation (Quattrone & Hopper, 2005). Nor does SAP inevitably revolutionise extant business practices. In another SAP implementation in a MNO (Quattrone & Hopper, 2005) transactions and business practices were left intact, albeit speeded up. Organisational participants resisted using SAP to integrate activities, collapse time and space, and increase central control for fear of upsetting extant power relations and controls. Transactions are defined by mediations and organisational struggles: they are not merely economic exchanges between predetermined entities. When a manager claims that SAP is a ‘‘great transaction processor’’ he neglects the long mediation that defined transactions, the entities involved, their constitutive relations, and how this remains ongoing. Accounting and SAP were intertwined abstract referencing systems built upon subjective choices and mediation, reciprocally modified by translations defining what was seen and how. Like Latour’s account of the Amazon Forest expedition, theories and technologies gave ontological status to an otherwise indistinguishable landmass. In Latour’s example the pedocompactor artefact expedited this. In the MNO accounting theories and techniques helped constitute the ontology of what was observable. Accounting (and SAP) did not produce static representations for managers continuously reinterpreted their demands during mediations. There was no consensus on how to achieve order: negotiating was time consuming and resolutions often proved temporary or unsatisfactory. Disputes made different information needs public and led users to reflect on their information needs and potential usages of SAP, which enabled them to continually discover information 236 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 needs and ways of utilising SAP unknown even to the consultants who sold SAP to the MNO. As a member of the IS Support Team at US HQ commented: The SAP product has evolved and our understanding of the product has evolved as well. We used to rely very heavily on consultants to help us understand all the features and flavours of SAP. Today I would say that there aren’t a whole lot of consultants out there that know a hell of a lot more than we do about SAP. Stating that every way of seeing is a way of not seeing may be trite but it informs how SAP’s features and functions were shaped. Ultimately, the Director of Accounting could ‘see’ subsidiaries dispersed across the globe and exert action at a distance from HQ. This visibility emerged continuously, putting the system in perpetual turmoil as multiple, changing centres of calculations and discretion emerged (see Munro, 1999; Quattrone & Hopper, 2001). SAP’s features could only be understood by tracing mediation in the spaces and times SAP intersected with and reinforced. Moreover, SAP’s incompleteness and its changing constitution enabled it to engage different constituencies simultaneously. It is the gap, the lacuna in the representation (the residual ‘other’ of classifications) that gives scope for change (Bowker & Leigh Star, 1999; Quattrone, 2006). Visibility and patterns of absence/presence were vital for creating management practices trying to clarify matters. They made some things present, be they transactions, departments, or behaviours, but they made other things absent. But absent objects are not necessarily passive, for what is made present and has impact relies on a process of ‘Othering’ that denotes absences (Law & Singleton, 2005, p. 343). Also absences can exist in organisational times and spaces elsewhere or lie dormant: they can ‘strike back’. This rendered SAP an object in constant turmoil and the rudimentary understanding of SAP’s potential persisted years later. The International Financial Controller in the EU HQ commented: Do we get more information and better reporting? I think we close the books quicker for sure. . . . You send the forecast to [HQ] much sooner. . . . When we close, we know the results are good, while before . . . no one could really check all of them at the same time. . . . SAP has all the information within its gut but . . . you need to know how to use it and how to extract it and there I don’t think we are far enough yet. . . . We know . . . some basics but it takes some time. A Budget Analyst in the USA HQ confirmed that: ‘‘We really need to get more comfortable with SAP. It is still relatively new for us. And we are finding a lot of new things we can do’’. Rather than being a package SAP was ‘boundary-less’ for its apparently clear features, functions, and architecture14 derived from continuous mediation and translation. Even four years after starting its implementation, when it appeared to operate smoothly, SAP was both ‘de-finite’ and ‘de-finite’, i.e. incomplete, as the UK Financial Controller recounted: [SAP] is almost evolutionary. It doesn’t stop. What typically happens is that you roll it out in a business and they make the first sort of beachhead and then they roll it out 14 SAP R/3 (release 3.0, the version used during the case study) has three main elements: a database that stores business transactions; various business modules, e.g. financial accounting and controlling, selling and distribution; and the basic software (or middleware) that links the database with the business modules (Hernández, 1997, pp. 15–71). P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 237 in another business and the lessons that they have learned . . . get transferred to the second business. . . .The trick is that as each business gets rolled out is to bring the businesses already on SAP up to the new level of knowledge. It’s actually quite difficult. SAP proved difficult to comprehend in detail and individual understandings changed with use. According to the Customer Service Leader of the UK subsidiary: [In] 1996, we went live with [SAP]. If somebody had presented me with a list of ‘‘This is what SAP will do for you in terms of management reports’’, Wow! You know. WOW! Here SAP was perceived as sufficiently malleable to accomplish many business requirements – some undiscovered. Hence some managers claimed that with SAP, ‘‘You can do about anything you want.’’ (IS Business Relationships Manager, USA HQ). SAP cannot be satisfactorily depicted as a stable black box for each de-finition contained germs of its de-finition. SAP expresses a theory, a way of seeing, that illuminates but also obscures organisational vistas. It helps managers make sense of events but also leaves space for different views. Thus every categorisation, every order, is open to question. It is impossible to escape from a progressive form: every order is intrinsically an ordering because it is partial and allows diversity (Quattrone, 2004b). SAP never reaches closure, hence never becomes a black box. The features and operational functions that define SAP are never finite because they invariably provoke multiple interpretations. If SAP were a product of de-finition (i.e. has clear features, parts, and functions to unambiguously meet user information needs) it would not travel well for it would loose the malleability necessary for mediating, forging alliances, engaging diversity, and satisfying users’ divergent and emergent information needs. Conceptualising SAP as a working technology with precise definitions (i.e. a well de-fined black box) neglects how each definition is ambiguous and incomplete, and how this enables SAP to translate, get translated, and engage a conglomerate of humans (users) and non-humans (e.g. accounting, excel spreadsheets). It is the tensions in ‘de-finitions’ that enable SAP to become a working IT system. 5. What is IT? SAP as a ‘heteromogeneous object’ 5.1. What is SAP? When the research began interviewees were asked: What does SAP do? What is it? As recounted above, some defined SAP as: ‘‘A superb transaction processor . . . very good at dealing with millions of transactions’’ (UK Financial Controller) and ‘‘A common database of information’’ (Director of Accounting in USA HQ). Managers across hierarchical levels, functions, and business divisions recounted how SAP made problems visible and easier to analyse. The stories below are typical: SAP helped us . . . to see the problem, to come up with a solution. It is giving me data. Instead of working on instinct, gut feel, ‘This looks right’, ‘That looks right’, I can actually quantify it and say ‘‘That is what is going on.’’ (UK Customer Service Leader) SAP will allow you to slice and dice however you want to see it. So if you want to see the consolidated corporation, you look at all companies, all business areas, and it all 238 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 goes together. If composites [a business division] wants to look at their global composites then they take all company codes and key on the business area and that gives them all composites. . . . They select certain company codes within the area and then that gives them composites Europe or composites North America or Brazil. . . . You can see what companies or what part of the world you are dealing with and so you can do different things (Accounting Specialist, USA HQ). After 4 years of mediations, customisations, ‘‘bolt-ons’’, and upgrades managers used SAP to quickly see what they wanted to in acceptable representations. Some managers perceived SAP as a homogeneous object. For example, the alliance of spreadsheets and SAP was so seamless that a Budget Analyst in the USA HQ continually referred to an accounting report generated by Excel as ‘SAP’. Now the local Manufacturing Controller could scrutinise costs in the Green Sheet line by line to ‘see’ events in his factory. European HQ Analysts in Brussels could examine aggregated ‘‘core measurement tools in the business’’ in SAP monthly performance accounts (products of mediations between themselves and their superiors). And, by following defined steps for consolidating financial data, HQ could construct vistas for the entire business, regions, subsidiaries, and plants. The Senior Accountant in USA HQ stated: From here we could run incomes, P& Ls, and balance sheets for the whole world, and we do basically. They [the subsidiaries]. . . and we can run the whole consolidation process here [pointing to the screen]. [The consolidation] could be broken up into steps. And I have visited a couple of these [subsidiaries] where they have broken this up – where one geographic region might run the consolidation for their piece, and then another area theirs, and another area theirs, and then overall consolidation is run centrally. So the system can be configured to do that. But for our purposes . . . we have been able to do all that from one central location for everybody. . . . If everything goes perfectly. . . you could run through . . . the whole consolidation process in . . . an hour. HQ accountants could now view a reference to a remote regional area on a computer screen. The Chief Executive can drill down to the point where he can see the cost of producing a role of insulation in China. He doesn’t need to go and phone someone and ask. He can, at his desk, go in and look . . . SAP is a powerful tool. (UK Plant Analyst) Paraphrasing Latour, the Chief Executive in the USA could: Thanks to inscriptions oversee and control a situation in which [he is] submerged. [He can] become superior to that which is greater than [him], and [he is] able to gather synoptically all the actions that occurred over many [years, necessary to refine SAP] and that [he] has since forgotten (Latour, 1999, p. 65). From a beautiful open space in the American corporate HQ, with views over the river, one could see individual business units through consolidated financial statements, and recombine this data for forecasting and budgeting. The Director of Accounting’s centrality reinforced his control over the entire process. The long process of producing order proved fruitful. Accounting numbers on computer screens now rendered activities dispersed globally as apparently homogeneous and comparable. SAP’s new order was symbolised in the spatial restructuring of the accounting department at HQ (see P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 239 Fig. 3). It rendered personnel as homogenous as the transactions they dealt with. After four years of mediations and translations SAP became an operational technology with apparently definite features that maintained linear communications between the centre and peripheries. Yet we argue that any homogeneity of SAP (and accounting representations it supplied) is spurious. It did not tally with observations elsewhere. SAP’s output did not become perceived as trustworthy and usable until it was mediated to accommodate requests from dispersed corners of the MNO. How the jungle of ‘‘millions of transactions’’ (UK Financial Controller) was transformed into manageable summary data resembles Latour’s description of the scientific expedition into the Amazon forest. The boundary between the forest and the savannah only existed once it was constituted on paper in a scientific report produced at the expedition’s end. Accounting references similarly define boundaries, for example between the centre and subsidiaries of MNOs. Aggregated accounting entries refer to, describe, and ultimately constitute attributes of ‘products’ and ‘cost centres’. When aggregation extends to the consolidated Profit & Loss account (P& L) and Balance Sheet then the ‘organisation’ as a whole is constituted. Thus SAP, allied to accounting references, created a particular notion of time and space (Carmona, Ezzamel, & Gutiérrez, 2002), shaped relations between entities, and enabled the MNO to comprehend, manage and reduce ‘distance’ between its centre and its periphery (Quattrone & Hopper, 2005). But this did not mean that SAP (or its accounting system) became black-boxed, for there was a continuous tension between maintaining it as a working technology and needs for mediation. The effort and resources to maintain an IT system that met users’ problems and emergent information needs meant a stable singular definition of SAP was illusory. The Director of Electronic and New Digital Technologies at USA HQ recounted how this was dealt with: If someone has an issue in the field, they have someone identified as a champion locally . . . If that champion can’t deal with it, they would contact a champion that is regional. We also have a limited number of IS resources in Europe . . . about twelve people over there. They serve as a second level support. So they might get engaged to say, ‘‘Is this a problem or is it not a problem?’’ . . . Once you recognise it and you realise that it is a system issue you call . . . our help desk. . . . We have somebody on support twenty-four hours a day. . . . So that goes to the primary resource person. If they don’t answer in 15 min, then they go to the secondary, and if they don’t answer in 15 min then they go back to the team leader, and if they don’t answer in 15 min, then it goes to the systems integration leader, and if they don’t answer in 15 min, that used to be me. . . . They never got to me, which was a good thing. . . . We have five development teams in place right now. One that focuses on finance, one that focuses on customer interactivity, one that focuses on drill management, one that focuses on manufacturing, and one that focuses on HR personnel issues even though we use Peoplesoft for that. This was apparent during upgrades for customisation impacted future mediations. New software releases by SAP’s supplier threatened negotiated orders from previous translations and mediations that reconfigured SAP, as a Corporate Budget Analyst reflected: 240 Fig. 3. The layout of the accounting area in the US HQ: The accounting eye looking at the organisational world. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 a P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 241 They knew [the new release of SAP] was going to be a problem. And we were very hesitant to go, because every time the next [version is out] you have to . . . test every bolt in the system to make sure it works. And if it doesn’t you . . . figure out what to do to make it fit. Customisation through ‘bolt-ons’ made adopting SAP upgrades difficult: We have been fairly slow to take on the updates . . . The new version comes out and you drop it in . . . and you have one hundred bolt-ons with all these customisations that you have to evaluate. So it is a . . . massive effort to upgrade (Financial Leader, USA HQ). ‘Bolt-ons’ – the material, practical products of mediation – on the one hand prevented SAP from collapsing but on the other, upgrades reopened conflicts to the point where the MNO even considered refusing to implement them. Again we see how SAP is part of a ‘community’ (de Laet & Mol, 2000): it requires collaboration with many constituencies with fluctuating influence. It was ‘a-centred’ having no core or periphery (Quattrone & Hopper, 2001). HQ may appear to control SAP but this is illusory for no-one totally commanded the system: ‘‘It is not the product of the eyes, the hands and the brain of a single man, but a result of collective action’’ (de Laet & Mol, 2000, p. 249). The order and visibility SAP produced could return to confusion at any point due to serendipity, human agency, external shocks, and inevitably incomplete representations. Thus SAP proliferates rather than homogenises organisational spaces and times (Jones et al., 2004; Quattrone & Hopper, 2001, 2005). 5.2. IT as a ‘heteromogeneous object’ SAP is an object because it is a ‘fact’. Jones et al. (2004) clarify this when addressing the etymology of organisations as objects: Organizations are ‘facts’ . . . not in the common understanding of a thing existing out there but in a sense that recovers the Latin etymology of the noun, since one of the sources of the word ‘fact’ is the verb facere (‘to make’). The word ‘fact’ also interestingly shares the etymology of the word ‘effect’ (from ex facere) and highlights the power of fact in affecting and influencing (from ad facere) the Other. . . . Although the actual (rather than the simply factual) connotation of the word ‘fact’ has been lost in the English language, it is still quite alive (although every day less and less) in other Latin-based languages (e.g. Italian). . . . There is nothing less factual than a ‘fact’, for the fact always depends on those actions that are performed in order for it to exist, but also on the passions that each constructed fact is able to generate in the Other. Thus, in order for [a] new identity to exist, it always needs to be made, accomplished, achieved (ibid, p. 733–734). This delineates necessary characteristics of an object. An object must be ‘‘done and enacted’’ (Mol, 1999, p. 77) to become fact (in the Latin meaning of factum – something made), which gives it multiple meanings. But where does this come about? An object’s ontology is diffused across a network that defines its nature and use. Material relations and managerial practices (e.g. delivering a product, calculating costs) that make and sustain objects (SAP here) give it a unity which is multiple because relationships 242 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 and practices are multifaceted.15 Mol (1999, p. 75) describes this as ‘‘ontological politics’’ whereby ‘‘the term politics, works to underline this active mode, this process of shaping, and the fact that its character is both open and contested’’. This is a necessary but insufficient condition for defining an object to incorporate the ‘mess’ Law and Singleton (2005) referred to. However, it does not resolve problematic dichotomies between stability and change, and homogeneity and heterogeneity – the focus of this paper. If enactment yields diversity, what brings stability? For the object to materialise – to become ‘fact’ – it must ‘‘make others do things’’ (Latour, ad vocem). Objects – as facts or ‘things’ – stabilise because they attract what is different from them, and they must continue to attract diversity. An object appears homogeneous because it can engage and mobilise actions in other ‘things’, whether people or technologies. An object’s combination of being a ‘fact’ and an attractor underpins our definition of heteromogeneous. Here an object is more than a social construction because it exists as if ‘fact’ (being done) but less than a realist tangible thing because it exists only in this network of action. The above summarises the model of ‘heteromogeneous objects’ in Fig. 1. Here SAP is created and translated through a continuous process of ‘de-fining’ information needs to mobilise and engage diversity through accounting and visibility. SAP represents ambiguous abstract terms such as ‘integration’, ‘common’; ‘global’ and ‘simple’ in their absence to entice interest and diversity. SAP, in its minimal configuration of an absent IT system, establishes presence because people refer to IT. This enables it to be an attractor that induces action from other things, notably translation and mediation that constitutes SAP as an object. In Latour’s words it is ‘‘an interface that becomes more and more describable when it learns to be affected by many more elements’’ (2004, p. 73). SAP is not a product of institutionalisation that homogenises objects and their usages for this excludes diverse behaviours that sustain SAP and can misleadingly label them as deviant. IT emerges by attracting diversity: these heterogeneous collaborations make SAP appear homogeneous. Hence we label the paradox as heteromogeneous. However, Fig. 1 does not delineate a linear evolution of SAP over time and space for each component of the model can act (or lie dormant) simultaneously in the same space, and later or never elsewhere. Time and space are not explanatory variables: IT’s emergence requires alternative spatiotemporal frameworks (Jones et al., 2004; Quattrone & Hopper, 2001, 2005). 6. Conclusions and future research implications This paper examined the ontology of IT drawing from observations of an ERP implementation in a large MNO and insights from ANT and STS (Dugdale, 1999; de Laet & Mol, 2000; Latour, 1999; Latour, 2005; Law, 2000; Law 2002; Law & Singleton, 2005; Mol, 1999). It concludes that IT systems are not stable black-boxed technological artefacts. Paradoxically, this establishes them as minimal actors that can extend and mobilise networks of relations with users and other information and cameral technologies (accounting here) to construct visibility. 15 See Dugdale, 1999, on the constitution of a committee as a uniform body comprising multiple interests; and Quattrone, 2004, on the constitution of the Jesuit Order as a continuous ordering attempt. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 243 SAP is mobilised during translation and engagement that reciprocally ‘de-fine’ users’ information needs and SAP’s features. IT is rendered an object and a working practice through continuous translation by human and non-human constituencies. This actor-network (Latour, 2005) is more than social construction for it includes technologies that acquire meaning and consistency through absences and presences. Treating IT as a heteromogeneous object addresses issues raised at the inception, i.e. IT’s relation to non linear change (Ciborra, 2000; Quattrone & Hopper, 2001); how relations between agency and structure affect IT systems’ evolution (e.g. Orlikowski, 2000); and how distinctions between change and stabilization create epistemological problems for defining and identifying IT in a spatio-temporal context (e.g. Robey et al., 2002). We do not (and cannot) define all the realities that IT can assume. These will vary according to how one studies IT, and it erroneously assumes they are finite and known. Privileging visibility and accounting issues develops one alternative to views that grant IT a fixed ontology. In this sense the aim of this paper is methodological. The first issue concerns lines of inquiry for studying managerial practices and IT using ANT and STS (see Jones, 1999; Walsham, 1997; Scott and Wagner, 2003). The paper demonstrates their potential and highlights some limitations, especially their explanation of the co-existence of stability and change, and heterogeneity and homogeneity. Research in IS and elsewhere has concentrated on the emergence of IT as black boxes. This helps clarify how IT systems acquire stability and become taken-for-granted but it neglects what happens when they achieve this status. This carries the danger of wrongly assuming that organisational worlds achieve order and stability once processes leading to black boxing are identified (often judged in terms of success or failure). However, in our case SAP changed continually and differently across various spaces and times, to meet emergent and latent demands whilst retaining an organisational identity. The dynamic between these two tensions is neglected. The notion of heteromogeneous object is offered as a step towards redressing this. The second and related point concerns translation. This is useful for understanding how organisational entities, spaces and times are created and are subject to constant mediation. However, the notion is a victim of its own success. Once a technology is established and black boxed, translation apparently disappears. Research fails to address how apparently stable and functioning managerial practices like an ERP or accounting system exist alongside continuous change in their functions (and/or their multiple interpretations). The paper, by focussing on relations between organisational practices like accounting, IT, and visibility hopefully makes some progress here but more work on combining notions of translations with a new ontology for IT and management practices is required. Finally, inscriptions are crucial for defining what is seen, and for explaining how IT systems foster action at a distance and disciplinary control (Elmes et al., 2005). However, this tends to be teleological and gives unitary depictions of actors’ interpretations of organisational events and features. In contrast, our case suggests that inscriptions, as made in accounting systems, can be interpreted differently whilst retaining sufficient common currency to mobilise concerted organisational action. The issue is not to describe how closure and totalising systems like action at a distance occur but rather theorising how inscriptions create centres of calculations and discretion, and why some inscriptions gain greater currency than others (Munro, 1999; Quattrone & Hopper, 2005). Research over-emphasises how universal interpretations of inscriptions (say an 244 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 accounting record or number on a SAP screen) homogenise behaviour and controls. Treating inscriptions as passive objects of interpretation stultifies understanding the symbolic and semiotic power of signs and symbols in contemporary societies, and how they actively engage others (Fabbri, 1998; Jones et al., 2004). The actions that various inscriptions (e.g. screens and ordered tables) prompt and the hope they instil in managers trying to create order in their chaotic daily lives are a poorly understood example. To conclude, although SAP may never become a ‘black-box’, the idea of heterogeneous object is useful for understanding how an IT system can be stable but in a continuous state of drift (Quattrone & Hopper, 2001). SAP’s ambiguous features and functionalities, the malleability granted by its programming language (i.e. the ABAP code), and its compatibility with other technologies (e.g. spreadsheets) and management practices (i.e., accounting) help make it a best seller. SAP and its designers intelligently transfer the complex task of defining its features (along with its cost and risks) to end users, who may believe they govern the process despite being submerged and overwhelmed by it. Users can gain the illusion that SAP’s technology has a scientific basis whereas its emergence resides in human creativity and mediation with people and non-human objects. At best SAP is an idea, an absent presence, which instils hopes of order in those who buy it – a belief that activities in MNOs can be integrated by making transactions visible and homogenous. How people define these concepts must be studied as a process, for people learn by doing (translating things in practice on a daily basis). SAP’s purveyors’ rhetoric about its ability to integrate activities plays to human inclinations to choose tangible, apparently objective, stand-alone packages as solutions. By so doing they manage to convince large MNOs to buy an idea, a myth which Solves All the Problems. Acknowledgement The project benefited from financial support from the Training and Mobility of Researchers Programme (TMR) of the European Commission through a Marie Curie Fellowship (Contract No. ERFMBICT983193), the Institute of Chartered Accountants in England & Wales (ICAEW, Contract No. 5-392), and CICYT (Spain, grant No. SEC 98-0282). Paolo Quattrone is also associated with the Department of Business and Social Studies of the University of Siena through the ‘Rientro dei Cervelli’ programme of the Italian Ministry of Education, whose support is greatly appreciated. The authors are grateful to the two anonymous referees, Chris Chapman, Jacques-Olivier Charron, Catalijna Coopmans, Barbara Czarniawska, Taka Fujioka, Jorge Gomes, John Law, Christine McLean, Annemarie Mol, Wanda Orlikowski, Daniel Robey, Steve Woolgar, and participants to the seminars in the Department of Accounting at HEC, Paris; and Warwick Business School for their comments. The usual disclaimers apply. Appendix 1. Interview schedule A. Background of the Organisation 1. How long have you been operating in the UK? 2. What was the reason for choosing UK for your operations? P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 245 3. How large is your organisation (e.g. in terms of annual sales, employees)? 4. What is the ownership status of your organisation? 5. What is the type of the manufacturing in your organisation (e.g. high-unite volume, low-unit value, job shop, etc.)? 6. What are the types of new manufacturing technology utilised in your organisation? 7. Please tell us about your product market share in Europe. 8. What are the major competitors in Europe? B. Orgaznisation’s goals 1. Please describe the major strategic goals and their recent changes. 2. Please describe the process by which organisational goal are decided, evaluated and modified. Please describe the involvement of headquarters and your subsidiaries in determining goals. 3. Are there any kind of conflict between the goals of the headquarters and subsidiaries? C. Headquarter/subsidiaries relationships 1. What kind of technology and/or knowledge is transferred from the headquarter to the subsidiary, and vice-versa? 2. Do local managers go to the US for formal and/or informal meetings and how often? 3. Do top US management visit your organisation? D. Management control systems From headquarter to subsidiaries 1. Please describe the management control system from headquarter to subsidiaries. 2. In your opinion, what does the headquarters consider to be the major purpose of the management accounting system? 3. What accounting reports do you send regularly to the headquarters? 4. Apart from the accounting department, is there any other department responsible for providing financial or non-financial information to headquarters? 5. What kind of accounting report do you send to headquarters? Within subsidiaries 1. Please describe the management control system at the subsidiary level. 2. In your opinion, what do you consider to be the major purpose of the management accounting system at the subsidiaries level? 3. A part from the accounting department, is there any other department responsible for providing financial or non-financial information to the management? 4. Do you receive any accounting report from headquarters? What do you do as the result of receiving them? E. Performance measures 1. What performance measures are used in your organisation? To who are they applied? Have they changed in recent years? 246 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 2. What is the role of headquarters in performance measures establishment? Are local managers consulted with respect to the measures that will be used to evaluate them? 3. Are they linked to the reward system? 4. How satisfied are you with the performance measurement system? 5. In your opinion, what does the organisation consider to be the major purpose of the performance measurement system? 6. Are performance results communicated to the responsible managers? 7. What are the types of financial and non financial indicators used in your organisation for setting goals? F. Changes 1. Have there been any significant changes in recent years in the company’s organisation and/or management style? Please describe them. 2. What effect have these changes had on the management control system? 3. In your opinion what changes should be made in your organisation? References Armstrong, P. (2002). Management, Image and Management Accounting. Critical Perspectives on Accounting, 13, 281–295. Barley, S. R. (1986). Technology as an occasion for structuring: evidence from observations of CT scanners and the social order of radiology departments. Administrative Science Quarterly, 31, 78–108. Bechky, B. A. (2003). Sharing meaning across occupational communities: the transformation of understanding on a production floor. Organization Science, 14(3), 312–330. Berger, P. L., & Luckmann, T. (1966). The social contruction of reality. New York: Doubleday and Co, The Penguin Press edition, 1967. Blain, J., Dodd, B. & Sandison, D. (1998). Administering SAP? R/3: The Fi-Financial Accounting and CoControlling Modules, ASAP World Consultancy, QUE Corporation, Indiana, IL. Bloomfield, B. P., & Vurdubakis, T. (1997). Vision of organization and organization of vision: the representational practices of information systems development. Accounting, Organizations and Society, 22(7), 639–668. Boudreau, M., & Robey, D. (2005). Enacting Integrated Information Technology: A Human Agency Perspective. Organization Science, 16(1), 3–18. Bowker, G., & Leigh Star, S. (1999). Sorting things out: classification and its consequences. Cambridge Mass: MIT Press. Briers, M., & Chua, W. F. (2001). The role of actor-networks and boundary objects in management accounting change: a field study of the implementation of activity-based costing. Accounting, Organizations and Society, 26, 237–269. Burns, J., & Scapens, R. W. (2000). Conceptualising management accounting change: an institutional framework. Management Accounting Research, 11, 3–25. Callon, M. (1981). Struggles and negotiation to define what is problematic and what is not: the sociology of translation. In K. D. Knorr R. Krhon & R. Whitley (Eds.), The Social Process of Scientific Investigation. Dordrecht-London: D. Reidel Publishing Company. Callon, M. (1986). The sociology of actor-network: the case of an electric vehicle. In: M. Callon, J. Law & A. Rip (Eds.), Mapping the dynamics of Science and Technology. Sociology of Science in the Real Word. Callon, M., & Latour, B. (1981). Unscrewing the big Leviathan: how actors macro-structure reality and how sociologists help them to do so. In K. Knorr-Cetina & A. V. Cicourel (Eds.), Advances in social theory and methodology (pp. 277–303). London: Routledge and Keegan Paul. Carlile, P. (2002). A pragmatic view of knowledge and boundaries: boundary objects in new product development. Organization Science, 13(4), 442–455. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 247 Carlile, P. (2004). Transferring, translating, and transforming: an interaive framework for making managing knowledge across boundaries. Organization Science, 15(5), 555–568. Carmona, S., Ezzamel, M., & Gutiérrez, F. (2002). The relationship between accounting and spatial practices in the factory. Accounting, Organizations and Society, 27(3), 239–274. Carmona, S., Gutiérrez, I., & Cámara, M. (1999). A profile of european accounting research. European Accounting Review, 8(3). Ciborra, C., (Ed.). (2000). From control to drift, the dynamics of corporate information infrastructures. Oxford University Press. Colós, S. (1996). What’s wrong with representation? testimonio and democratic culture, 1995. In George M. Gugelberger (Ed.), The real thing. testimonial discourse and Latin America (pp. 161–171). London: Duke University Press. Czarniawska, B. (1998). Narrative approaches in organisation studies. Thousand Oaks, London: Sage. Czarniawska, B. (1999). Writing management. Organisation theory as literary genre. Oxford: Oxford University Press. Czarniawska, B. (2004). On time, space, and action nets. Organization, 11(6). Dechow, N., & Mouritsen, J. (2005). Enterprise resource planning systems, management control and the quest for integration. Accounting, Organizations and Society, 30(7–8), 691–733. de Laet, M., & Mol, A. (2000). The Zimbabwe bush pump: mechanics of a fluid technology. Social Studies of Science, 30(2), 225–263. DeSanctis, G., & Poole, M. S. (1994). Capturing the complexity in advanced technology use: adaptive structuration theory. Organization Science, 5(2), 121–147. Dugdale, A. (1999). Materiality: juggling sameness and difference. In J. Law & J. Hassard (Eds.), Actor network theories and after (pp. 113–135). London: Sage. EAR (2003). European accounting review, 2003. Special issue on management control and new information technologies, 12(112), 1. Elmes, M. B., Strong, D. M., & Volkoff, O. (2005). Panoptic empowerment and reflective conformity in enterprise systems-enabled organizations. Information and Organization, 15, 1–37. Fabbri, P. (1998). La svolta semiotica. Bari: Laterza. Fujimura, J. (1992). Crafting science: standardized packages, boundary objects, and ‘‘translation’’’. In A. Pickering (Ed.), Science as practice and culture (pp. 168–211). Chicago: University of Chicago Press. Giddens, A. (1989). A reply to my critiquesSocial Theory and Modern Society (pp. 249–301). Cambridge: David Held/John B. Thompson. Granlund, M., & Malmi, T. (2002). Moderate impact of ERPS on management accounting: a lag or permanent outcome? Management Accounting Research, 13, 299–321. Hansen, A., & Mouritsen, J. (1999). Managerial technology and netted networks. ’competitiveness’ in action: the work of translating performance in a high-tech firm. Organization, 6(3), 451–471. Hernández, J. A. (1997). The SAP R/3 Handbook. New York: Computing McGraw Hill. Hines, R. (1988). Financial accounting. In communicating reality we construct reality. Accounting, Organizations and Society, 13(3), 251–261. Hofstadter, D. R. (1979). Gödel, Escher, Bach: An eternal golden braid. Basic Books. Hopwood, A. G. (1980). Organizational and behavioural aspects of budgeting and control. In J. Arnold B. Carsberg & R. Scapens (Eds.), Topics in management accounting (pp. 221–240). Prentice Hall. Jones, M. (1999). Information systems and the double mangle: steering a course between the Scylla of embedded structure and the Charybdis of strong asymmetry. In T. J. Larsen L. Levine & J. I. DeGross (Eds.), Information systems: current issues and future changes (pp. 87–306). Laxemburg: IFIP. Jones, G., Mclean, C., & Quattrone, P. (2004). Spacing and Timing. Introduction to the Special Issue of Organization on ‘Spacing and Timing’, 11(6), 723–741. Kallinikos, J. (2004). Deconstructing information packages. Organizational and behavioural implication of ERP systems. Information technology and People, 17(1), 8–30. Latour, B. (1986). The powers of association. In Law, J., (Ed.). Power, action, and belief: A new sociology of knowledge. Sociological Review Monographs (vol. 32) (pp. 264–280). London: Routledge. Latour, B. (1987). Science in action. How to follow scientist and engineers through society. Cambridge Mass: Harvard University Press. 248 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 Latour, B. (1988). The politics of explanations: An alternative. In S. Woolgar (Ed.), Knowledge and reflexivity. new frontiers in the sociology of knowledge. London: Sage Publications. Latour, B. (1991). We have never been modern. London: Sage Publications. Latour, B. (1997). Trains of thought-piaget, formalism and the fifth dimension. Common Knowledge, 3(6), 170–191. Latour, B. (1999). How to talk about the body? The normative dimension of science studies. Available from http://www.ensmp.fr/~latour/articles/1999.html. Latour, B. (2001). Gabriel tarde and the end of the social. In Joyce Patrick (Ed.), The social in question. New bearings in history and the social sciences (pp. 17–132). London: Routledge. Latour, B. (2002). How to talk about the body? The normative dimension of science studies. Body and Society. Available from http://www.ensmp.fr/~latour/articles/index.html. Latour, B. (2005). Reassembling the social. Oxford: Oxford University Press. Law, J. (2000a). Objects, spaces and others. Published by the Centre for Science Studies, Lancaster University, Lancaster, LA14YN, UK. Available from http://www.lancs.ac.uk/fss/sociology/papers/law-objects-spacesothers.pdf. Law, J. (2000b). Objects, spaces and others. Published by the Centre for Science Studies, Lancaster University, Lancaster, LA14YN, UK. Available from http://www.lancs.ac.uk/fss/sociology/papers/law-economics-asinterference.pdf. Law, J., & Hassard, J. (Eds.). (1999). Actor network theories and after. London: Sage. Law, J., & Singleton, V. (2005). Object Lessons. Organization, 12(3), 331–355. Macintosh, N. B. (2002). Accounting, accountants and accountability. Poststructuralist positions. London: Routledge. Mackenzie, D. (2003). These things called systems: Collective Imaginings and infrastructural software. Social Studies of Science, 33(3), 365–387. Markus, L. M., & Robey, D. (1988). Information technology and organizational change: causal structure in theory and research. Management Science, 35(5), 583–598. McCloskey, D. N. (1985). Rethoric of economics. Madison: The University of Winsconsin Press. Miller, P., & Napier, C. (1993). Genealogies of calculations, accounting. Organizations and Society, 18(7/8), 631–647. Mol, A. (1999). Ontological Politics. A word and some questions. In J. Law & J. Hassard (Eds.), Actor Network Theories and After. London: Sage. Morin, E. (1977), La méthode. I. La nature de la nature, éditions du Seuil, Paris (It. trans., Il metodo, Feltrinelli, Milan, 1983). Mouritsen, J. (2005). Beyond accounting change: design and mobilization of management control systems. Journal of Accounting and Organizational Change, 1(1), 97–113. Munro, R. (1999). Power and discretion: membership work in the time of technology. Organization, 6(3). Newell, S., Huang, J. C., Galliers, R. D., & Pan, S. L. (2003). Implementing enterprise resource planning and knowledge management systems in tandem: fostering efficiency and innovation complementarity. Information and Organization, 13(1), 25–52. Organization, (2005). Special issue on ‘Objects’ 12, 3. Orlikowski, W. (1992). The duality of technology: rethinking the concept of technology in organizations. Organization Science, 3(3), 398–427. Orlikowski, W. (1996). Improvising Organisational transformation over time: a situated change perspective. Information Systems Research, 7(1), 63–92. Orlikowski, W. (2000). Using technology and constituting structures: a practice lens for studying technology in organizations. Organization Science, 11(4), 404–428. Orlikowski, W., & Barley, S. R. (2001). Technology and institutions: what can research on information technology and research on organizations learn from each other? MIS quarterly, 25(2), 145–165. Orlikowski, W., & Iacono, C. S. (2001). Research commentary: desperately seeking the ‘‘IT’’ in IT research – a call to theorizing the IT artifact. Information Systems Research, 12(2), 121–134. Orlikowski, W., & Robey, D. (1991). Information theory and the structuring of organizations. Information Systems Research, 2(2), 143–169. Orlikowski, W., & Yates, J. (2002). It’s about time: temporal structuring in organizations. Organization Science, 13(6), 684–700. P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 249 Pawlowski, S. D., & Robey, D. (2004). Bridging user organizations: knowledge brokering and the work of information technology professionals. MIS quarterly, 28(4), 645–672. Poole, M. S. & DeSanctis, G. (2004). Structuration Theory in information systems research: method and controversies. In: M. E. Whitman & A. B. Woszczynski (Eds.), The handbook of information systems research (pp. 206–249). Quattrone, P. (2004a). Commenting on a commentary? Methodological Choices in Accounting Research. Critical Perspectives on Accounting, 15(2), 232–247. Quattrone, P. (2004b). Accounting for god. Accounting and accountability practices in the society of Jesus (Italy, 16th–17th centuries). Accounting, Organizations and Society, 29(7), 647–683. Quattrone, P. (2006). The possibility of the testimony. A case for case study research. Organization, 13(1), 143–157. Quattrone, P., & Hopper, T. (2001). What does organisational change mean? Speculations on a Taken-for-granted Category, Management Accounting Research, 12(4), 403–435. Quattrone, P., & Hopper, T. (2005). A ‘Time-space odyssey’: management control systems in multinational organisations. Accounting, Organizations and Society, 30(7–8), 735–764. Robey, D., & Boudreau, M. (1999). Accounting for the contradictory organizational consequences of information technology: theoretical directions and methodological implications. Information Systems Research, 10(2), 167–197. Robey, D., Ross, J. W., & Boudreau, M. (2002). Learning to implement enterprise systems: an explanatory study of the dialectics of change. Journal of Management Information systems, 19(1), 17–46. Robey, D., & Sahay, S. (1996). Transforming work through information technology: a comparative case study of geographic information systems in county government. Information Systems Research, 7(1), 93–110. Robson, K. (1991). On the arenas of accounting change: the process of translation accounting. Organisations and Society, 16(3), 547–570. Robson, K. (1992). Accounting numbers as ‘‘inscription’’: action at a distance and the development of accounting. Accounting Organisations and Society, 17(7), 685. Scapens, R. W. & Jazayeri, M. (1999). The local implementation of a global ERP system, working paper. Schultze, U., & Orlikowski, W. (2004). A practice perspective on technology-mediated network relations: the use of internet-based self-serve technology. Information Systems Research, 15(1), 87–106. Scott, S. V., & Wagner, E. L. (2003). Networks, negotiations, and new times: The implementation of enterprise resource planning into an academic administration. Information and Organization, 13(4), 285–313. Shryrer, C. F. (1993). Records as Genres. Written Comm., 10, 200–234. Star, S. L., & Griesemer, J. R. (1989). Institutional ecology, ‘translations’ and boundary objects: amateurs and professionals in Berkeley’s museum of vertebrate zoology. Social Studies of Science, 19, 387–420. Stone, D. N. (2001). ‘Accountant’s tales’, accounting. Organizations and Society, 26, 461–470. Suzuki, T. (2003). The espistemology of Macroeconomic reality: The Keynesian revolution from an accounting point of view. Accounting, Organizations and Society, 28(7), 471–517. Tarde, G. (1999). Monadologie et sociologie. Paris: Les empêcheurs de penseren rond. Thompson, J. (1989). The theory of structuration. In D. Held & J. Thompson (Eds.), Social theory of modern societies: Anthony Giddens and his critics (pp. 56–76). Cambridge, UK: Cambridge University Press. Varela, F. J., Thompson, E., & Rosch, E. (1991). The embodied mind, cognitive science and human experience. Massachusetts Institute of Technology Press. Walsham, G. (1997). Actor-network theory and IS research: current status and future prospects. In A. S. Lee J. Liebenau & J. I. DeGross (Eds.), Information systems and qualitative research (pp. 466–480). New York: Chapman & Hall. Whitman, M., & Woszczynski, A. (Eds.). (2004). The handbook of information systems research. Hershey, PA: Idea Group Publishing. Woolgar, S. (1981). Discovery: logic and sequence in a scientific text. In K. D. Knorr R. Krhon & R. Whitley (Eds.), The social process of scientific investigation. Dordrecht-London: D. Reidel Publishing Company. Woolgar, S. (1988). Science: The very idea. Sussex: Ellis Horwood. 250 P. Quattrone, T. Hopper / Information and Organization 16 (2006) 212–250 Woolgar, S. (1991). Configuring the user: the case of usability trials. In J. Law (Ed.), A sociology of monsters: essays on power, techonology and domination. London and New York: Routledge. Young, J. J. (2003). Constructing, persuading and silencing: the rhetoric of accounting standards. Accounting, Organizations and Society, 28(6), 621–638.