The Techno-Political Dynamics of Information Infrastructure Development: Interpreting Two Cases of Puzzling Evidence

Transcription

1 The Techno-Political Dynamics of Information Infrastructure Development: Interpreting Two Cases of Puzzling Evidence Knut H. Rolland Department of Computer and Information Science, Norwegian University of Science and Technology, Trondheim, Norway Margunn Aanestad Department of Informatics, University of Oslo, Oslo, Norway Abstract. This paper describes a preliminary analysis of the role of power in the development of two large-scale information infrastructures. Grounded in these empirical studies, the paper suggests that such largescale infrastructure efforts encapsulate a specific techno-political dynamics that by necessity involve the need to juggle a variety of changing alliances, to give and take, exercise power more or less openly, and to inscribe ways of disciplining different users patterns of use. Consequently, in contrast to much written in the IS literature on the issue of power, this paper argues that power is not necessarily negative, stable, and zero-sum, but rather performed, dialectic, distributed, implicit, and inscribed. The paper concludes by sketching some ideas for conceptualizing the role of power in the development and use of information infrastructures and such mega-projects in general.

2 Keywords: Information infrastructure development, Power and IT, Integration 1 Introduction It has almost become a cliché to argue that information and communication technologies in organizations and society cannot be understood solely according to their technical features and components. Information infrastructures are relatively open-ended and are often extended and used for different and changing purposes by different user groups (e.g. Ciborra et al., 2000; Star and Ruhleder, 1996). Thus, like the more traditional infrastructures such as roads and transport systems, their use is regulated by formal traffic rules, traffic signs and more informal institutionalized practices. Without taking this analogy too far it is obvious that corporate information infrastructures and global infrastructures like the Internet also involve a range of different technical as well as non-technical means for regulating users behaviour. Thus, design of information infrastructures as the back-bone of the information society is inherently politically charged. One example of the politics of infrastructures is vividly illustrated in Abbate s famous study of the battle between the telecom companies who supported the X.25 protocol and those who supported Internet standards (Abbate, 1994). It is therefore hard to argue against the need for studying information infrastructures with a political lens as such technologies cannot be seen as independent from different coalitions pursuing different interests and the ongoing negotiations and battles between different actors that inevitably shape both development and use of infrastructures. In addition, the recent development and diffusion of integration technologies such as XML, Web services, various middleware platforms (e.g. COM, CORBA), and off-the-shelf software packages (i.e. ERP, CRM) encapsulates a considerable number of alternatives for developing extensive information infrastructures. Consequently, both within corporations and in the public sector these technologies seem to have enabled and legitimized increasingly bigger and more ambitious projects. While previous empirical studies have documented how mega-projects tend adopt a specific dynamics involving path-dependencies, lockins, and the globalization of side effects (Ciborra et al., 2000; Hanseth, Ciborra, and Braa, 2002; Monteiro, 1998) as a consequence of large-scale integration, this paper argues that large-scale integration encapsulate a subtle techno-political dynamics. We argue that the nature of this dynamism is closely related to how such infrastructure projects at the one hand need to be big and all-embracing in terms of number of users and functionality in order to mobilize the relevant actors up front (e.g. Ellingsen and Monteiro, 2001), but on the other hand, if mobilization is successful, this makes such technologies and their visions politically problematic to change.

3 The empirical basis for this argument is based on a study of the techno-political dynamics in two major information infrastructure initiatives in a global company and in the public health care sector respectively. However, power is a complex and rather ambiguous issue in itself that can be defined and theoretically conceptualized in multiple ways. In order to discuss the role of power in shaping the use of information infrastructures then, we will first discuss the concept of power as it has been perceived from a range of different perspectives in relation to technologies in general and IS in particular. Then, we will present a brief historical outline of the two case studies and a tentative interpretation of these cases through a theoretical lens viewing power as performed, dialectic, distributed, implicit, and inscribed. Then in the final section, we will suggest a conceptualization of the role of power in information infrastructures grounded in Science and Technology Studies (STS) that has lately become quite popular in IS research particularly a direction within STS referred to as actor-network theory (ANT). ANT combined with Foucault s notion of disciplinary power, we will argue, is quite promising as a way of analyzing the more systemic and implicit aspects of power in relation to use of information infrastructures and information and communication technologies in general. 2 Power and politics in is research and its theoretical grounding 2.1 The issue of power tackled briefly In the literature that considers the issue of power in relation to technology, one can contrast two different extreme view points. Firstly, in an orthodox engineering perspective technologies would be viewed as neutral and hence from this perspective power is invisible: it simply doesn t exist (Clegg and Wilson, 1991). Secondly, at the other extreme, in an orthodox labour process approach power would always be linked to capital and perceived as zero-sum. Thus, technology is the medium through which power is exercised and it is always zero-sum and deterministic in the sense that technologies have predictable outcomes and effects. An example of this is Braverman s seminal text Labour and Monopoly capital (1974) there power is linked to technology through the concept of control. Thus these views can be argued to both be flawed since they portray power and politics as either negative or non-existing. For instance in relation to the introduction of information technologies in organizations there is often talk about how design decisions were influenced by politics, the implementation of an information system failed because of politics, and that conservative users resists new technology. Moreover

4 in perceived success stories about information technologies such as American Airlines SABRE system, issues of power is totally invisible. In contrast to such assumptions, social theorists like Giddens and Foucault sees power as intrinsic in all human activity and actions. Giddens (1984) outlines an understanding of social action where action logically involves power in the sense of transformative capacity (p. 15). Giddens avoids the rather deterministic view of power found in classical structuralist school organization theory (e.g. Weber, 1947; Dahl, 1957) and earlier social theory, by conceptualizing structures of domination such as formal authority as both enabling and constraining. According to this framework, social actions involve structures of domination such as formal authority that are produced and re-produced in social actions, but do not necessarily determine human actions. Giddens (1984: p. 16) argues that all forms of dependence offer some resources whereby those who are subordinate can influence the activities of their superiors. And this is by Giddens referred to as the dialectic of control. Furthermore, along these lines power is not necessarily linked to conflict and power is not inherently oppressive. Giddens (1984: p. 257) defines power in the following way (quoted in Walsham, 1993: p.39): Power is the capacity to achieve outcomes; whether or not these are connected to purely sectional interests is not germane to its definition. Power is not, as such, an obstacle to freedom or emancipation but is their very medium although it would be foolish, of course, to ignore its constraining consequences. A different perspective which is relevant for understanding the role of power in understanding the dynamics of information infrastructures is Foucault s notion of disciplinary power (Foucault, 1972, 1977, 1984). According to Foucault power is embodied in heterogeneous micro-practices and power is seen as enacted and discontinuous rather than stable and exercised by a central actor (Thompson and McHugh, 1995). In this perspective power is also much more implicit and systemic in the sense that power is embedded and institutionalized in a myriad of different ways. For instance, whereas power in traditional terms where power had depended on personalised bonds of obligation, disciplinary power were developed and refined in networks of religious institutions, prisons, asylums, hospitals and workhouses at a local level. Likewise, in modern organizations power is embedded in a range of different technologies, best-practices, formal policies, management ideologies, professional codes, and work practices rather than explicitly exercised from a centralized actor with all-embracing control and power.

5 2.2 The issue of power in IS research Broadly one can distinguish IS studies that look at use of information and communication technologies from a political perspective into at least three different categories. Firstly, we have an early theoretical stream of IS studies drawing from the power school of OT that analyzed use of computers, IT-based organizational transformation and information systems development from the perspective of organizational politics (e.g. Kling, 1980; Markus, 1983; Robey and Markus, 1984). A salient point in this literature is that use and organizational consequences of computers cannot be understood in isolation from a wider context of social relations and henceforth this literature argues that the theoretical perspective of organizational politics are more appropriate for describing the dynamics of information technologies in organizational settings than rational models of computing. According to Kling (1980) whereas the rational models describe organizations as unified and having consistent and common goals, the political model emphasises that different individuals and groups typically have different interests (Which much in line with the OT of Shafritz and Ott, 1992 and Mintzberg (1983). Thus, accepted technologies are always technologies that serve specific interests (Kling, 1980). This political metaphor of IS has also been perceived as valuable in analysing empirical studies. For example, a seminal article by Markus (1983) power, politics and MIS implementation draws from the political perspective of IS implementation in order to account for user resistance of a financial information system. Markus (1983) highlights a conflict between different actors in the organization and explains user resistance to the new technology as a result of the specific interaction between the distribution of power and the features of the system. Another well-known empirical study is the study of Kling and Iacono (1984) who focus on how different interests groups try to influence an information system in an organization referred to as PRINTCO after implementation. Kling and Iacono find that key participants who value particular CBIS configurations actively strive to develop and expand them through a variety of strategies which require political mobilisation (p. 1218). In this case actors draw upon both structural and ideological aspects in order to mobilise support for a specific trajectory. For example, some actors used what the authors refer to as the language of efficiency to push the IS in a direction that increased their own power and control in the organization. These studies very well illustrate the use of an organizational politics perspective in IS research, but do neither discuss how power is enacted by different actors over time nor do they emphasize the dialectics of control or the more indirect forms of disciplinary power. Secondly, out of Scandinavia in the early seventies came a direction in IS research that more than anything explicitly focused on the politics of introducing technologies into the work place. The origin for what is often referred to the trade unionist or the

6 Scandinavian school in IS came from a project run by Kristen Nygaard and colleagues at the Norwegian computing centre. The project was initiated by the Norwegian Iron and Metal Workers union and started in Nygaard and his colleagues aimed at strengthening the workers bargaining position regarding introduction and use of new technologies and the project finally resulted in the technology agreements between workers union and the employers union and started a trend of politically inspired projects around Scandinavia (Iivari and Lyytinen, 1998). The basic assumption in many of these projects was that information technologies contributed to deskilling of workers and an unequal distribution of power between managers and workers in organizations (Bjerknes and Bratteteig, 1995), and furthermore that use of information technologies was inevitably related to a power struggle between workers and management. Thus, in this sense this literature seems to lean on a traditional perspective on power as something which some groups have some groups have not. Whereas these early projects took an active role in choosing to support resource week groups and expressed radical Marxist tones and goals, later work on participatory design in the 1980s and 1990s has drawn more on a tool view of IT grounded in the philosophies of Heidegger and Wittgenstein, and focused more on developing a range of different techniques such as mock-ups, language games and collaborative prototyping in order to ensure user participation (Iivari and Lyytinen, 1998). Although PD school has succeeded in multiple ways in introducing new techniques, introduced laws that ensure user participation in planning and implementation of new technologies, and the emphasized the importance of user participation in ISD, empirical research in Norway show that in practice users have little impact on ISD and implementation of information technologies (Hatling and Sørensen, 1998). The reason for this is as argued by Hatling and Sørensen that software designers and managers typically interpret user participation in different ways and tend to perceive PD as merely a set of techniques. Thus, as practiced in industry, PD seems to have lost its initial political perspective, and it is maybe a great paradox that many of these techniques have now been incorporated into consultant techniques simply to improve control over requirements (and thereby users). In addition I will argue that conflicts and power in relation to development and use of information and communication technologies in contemporary organizations does not simply reflect a class struggle, and hence PD has not analytically kept up with the changes in technologies, global markets, more complex trade unions. A third category of is research that adopts a political perspective has drawn from social theorists like Giddens and Foucault. This literature tends to treat the issue of power in more depth and acknowledges that power endemic to human action and is not necessarily viewed as negative. Interesting examples of such research is Orlikowski s article called integrated information environment or matrix of control and draws

7 from both Giddens and Foucault in order to emphasize the systemic nature of power as well as the dialectics of control in relation to use of a CASE tool in an IT consultant company. Among other things, Orlikowski concludes that IT often tends to reproduce and strengthen existing distributions of power. In his book Interpreting Information Systems in Organizations Walsham (1993) draws from Morgan s political metaphor of organizations and Giddens structuration theory in order to understand how information systems shapes and are shaped in a wider organizational context. This early work on power within Information Systems Research outlines a more realistic perspective on the interplay between IT and organization than the purely normative and technical perspectives. However, due to the later changes in work life and organizations these theorizations are also too coarse for accounting for the role of power in development and use of IT in general and information infrastructures in particular. In particular we disagree with the static view of power often found in the IS literature implying that information and communication technologies serve some specific actors interests (e.g. managers) and imply a tighter control over other actors (e.g. users). Even when hierarchical power is executed, the actions taken to implicitly and explicitly discipline use of an information infrastructure may have unintended side effects. These again may lead to lack of control. Thus, we argue that power should rather be seen as always dynamic, distributed, partial and performed. 3 Case # 1: The telemedicine project Here we will describe a partly successful telemedicine project, where a broadband network was set up between two major Norwegian hospitals. This project has been the main case for the PhD work of one of the authors. In other publications, several aspects of the process has been analyzed, for example the non-planned nature of the processes (Johansen et al, 1999; Aanestad and Hanseth, 2000), how the characteristics of the technology influenced the processes (Aanestad 2003) and the way learning occurred (Aanestad and Hanseth, 2001; Aanestad, 2002), while in this paper, the focus will be on examining the techno-political nature of the process. As telecommunication deregulation was approaching in Norway (by ), the Swedish telecom Telia approached prospective customers in Norway. An initial demonstration convinced the two hospitals that the broadband (ATM) technology opened up possibilities for developing a wide range of new medical services related to minimal invasive surgery. A project entitled Development of Interactive Medical Services (DIMedS) was established in 1998 between Telia, Ericsson, Rikshospitalet, Ullevål hospital and the University of Oslo. The intended usage was to transmit live audio and video for demonstrations of surgical procedures to remotely located surgeons, and thus to rationalize the long specialist training. The possibility to discuss

8 specific patient cases and emergent problems with other experts could also be utilized, as well as recording and editing the video with the aim of offering video server functionality. 3.1 The institutional and political context of the project Ullevål hospital has a history of ambitious and innovative ICT-projects, and during this time period ( ) the health region of Eastern Norway, where Ullevål was the regional hospital, were about to choose a vendor for a regional broadband network to connect several health care institutions at different levels. Two of the major candidates were Norwegian Telenor and Swedish Telia. While the region eventually chose Telenor, Ullevål was simultaneously connected to Telia s ATM network through this project. However, the DIMedS project agreement stated that transmissions within this project should not be forwarded to other competing networks, in practice the regional Telenor network. At some occasions other transmission technologies (ISDN, satellite) were used to expand the reach of the DIMedS network. Ullevål and Rikshospitalet are two of the largest hospitals in Norway, both located in central Oslo. While previously they were well known for their competitive attitude to each other, the weight of the expertise located here constitutes the main share of surgical expertise in Norway. However, in 1996 a National Center for Advanced Laparoscopic Surgery was located in Trondheim and not in Oslo, partly due to governmental policy of supporting regional development outside the capital. This center offered courses for surgeons specializing in laparoscopy and a follow-up telemedicine service for surgeons. The newly trained surgeons could have an expert follow them through their five first operations when they were back on their ordinary work place. The technology used for this was low cost ISDN videoconferencing at 384 kbits/s. According to a sketch for the DIMedS project plan (Wörkvist, 1998), one of the long term project goals were to establish a telemedicine training center in Oslo. The rhetoric of the Oslo-based DIMedS project implied a dismissal of the ISDN technology as being inadequate for such complex and critical procedures as laparoscopic surgery (Røtnes et al, 1998, Buanes et al., 1999). A much used PowerPoint slide displayed two versions of a single video frame (image) from a surgical procedure. On the one transmitted via the ATM network a tiny bleeding was visible, while this could not be seen on the frame that had been compressed more heavily and transmitted on the ISDN network. The concept of the necessary high image quality, with such evidential material as this image, was given a central role at this point of the project and the presentations at national and international conferences. Apart from being established separately from the national competence and training centre in Trondheim, the project was also not formally associated with the National Center for Telemedicine, which had been established in Tromsø in This center

9 had been developing applications and services for a mass market, much for primary health care, using low cost technologies like telephone line transmissions (analogue lines and ISDN). There had not been much activity towards the goals of the DIMedS project, i.e. providing high end services for the specialist groups using front end technologies. The national center attracted most of the public funding for telemedicine development. The DIMedS project did not receive public funding, but was carried out through the support of the industry partners when it concerned network access and lending equipment. Within the hospitals parts of the equipment were already in place. In addition small economic resources were identified and used for purchasing additional equipment. The work was carried out in the time were able to free from other tasks, and the activities depended heavily on the services of draft people (conscientious objectors). There were thus a spirit of self-reliance; of carrying through the project despite the lack of support and financing from the public government. 3.2 Ambitions and their adjustments The combined weight of the hospitals professional importance and their geographical location (implying high access to patients) did certainly play a role in Telia s and Ericsson s decision to support the project. The project aims and visions were quite ambitious. One of the scenarios that were discussed involved reducing the time of specialist education in Norway, from the current years on average to a 4 6 year period. If this could be possible, it is obvious that the benefit for Norwegian health care would be significant. This was to be achieved through offering interactive access to live operations, and in a later stage, to recorded and edited video material. For Telia, the DIMedS project was not intended to be a commercial product development product, but rather a market development project. After the implementation of the broadband network technology in September 1998, the work to generate relevant usage of the system started. Initially many technical tests and proof of concept -transmissions were carried out. When the suitability of the technology had been verified, i.e. that the surgeons were satisfied with the image quality, the industry partners expected the hospital partners to generate higher amounts of real usage. Then the project encountered difficulties with summoning the intended receivers (surgeons) at given points of time. It was difficult for the surgeons at the receiving site to change the work schedules to accommodate the transmission, in particular since it was not included in any formal training programme and did not give credit points. The result of this was that there were not many surgeons at Ullevål that were able or interested in watching transmissions from Rikshospitalet. Some surgeons in the Eastern Norway region expressed interest in watching, but the contract precluded a direct coupling between the Telia and the Telenor networks. There were not other

10 hospitals where this technology was implemented, so no other communication partners existed. Thus, in order to utilize the network access and video digitizing equipment fully, the scope of the project s activities was expanded. In order to enroll other user groups within the hospitals in the telemedicine activities, the project went beyond the surgical education plans and offered to arrange several sessions that would demonstrate the technical possibilities. At one instance the large weekly staff meeting at Rikshospitalet was conducted on-line with Ullevål hospital s staff meeting, with some speakers at each site. As the two hospitals were well known for their competitive attitude towards each other, this was seen as remarkable and attracted quite a lot of attention. The session was also transmitted to a nearby hotel hosting the yearly national telemedicine conference. Later other high-profile transmissions as well as smaller transmissions were conducted, ranging from whole-day regional seminars to half an hour s lunch meetings. Elsewhere this process has been analyzed and it is argued that these activities provided excellent opportunities for learning to handle and utilize the technology (Aanestad and Hanseth, 2001). However, the actions in this phase also shaped the project and the resulting technical setup of the departments significantly. Specific pieces of equipment were borrowed or purchased for these purposes. For example could the client departments in the hospitals buy minor devices that were needed (e.g. an extra microphone), which afterwards were donated to the project. Also from other sources individual pieces of equipment was donated, usually in relation to specific transmissions. The resulting portfolio of equipment again determined which use areas were possible to serve (i.e. what kind of transmission could be handled). 3.3 Results At this time the Interventional Centre was located in temporary premises while the new Rikshospitalet was planned and built. This precluded large investments in technical infrastructure that would not be moved. The move date was repeatedly postponed, and the time span of the first sub-project in the DIMedS project had to be renegotiated. When this sub-project ended in June 1999 no further sub-projects were initiated, as the date for the actual move was still undecided (it eventually occurred in May 2000). The overall DIMedS project was discontinued in the autumn of 1999 even though the initial plans were a three years project, i.e. until The industry partners hesitation to continue the project may have several explanations beyond the postponed move, but one significant factor at this point in time was an expected merger between the Norwegian and the Swedish telcos, Telenor and Telia (the project partner). This led to postponing of strategic decisions regarding telemedicine activities within Telia. Although the attempted merger eventually failed (autumn 1999), the project was not reactivated.

11 At that time no alternative broadband connection was in place, and the telemedicine activities were restricted to occasional ISDN transmissions. After the DIMedS project the control room functioned as the technological hub for external and internal audio/video transmissions for the whole Rikshospitalet, and several departments were using the facilities frequently, e.g. to transmit from the local operation theatres to meeting rooms or lecture halls. In that respect the project work can be said to have been successful. But it seems appropriate to say that the project failed in establishing a market for broadband-based educational services for surgeons in Norway. The visions that were produced in order to enroll support never materialized. These visions were too grand and required establishing new structures (e.g. educational programmes) that were beyond the scope of the actors and the projects. Nevertheless, these grand ambitions were a significant factor in the process, as they served to mobilize support (see also Ellingsen and Monteiro, 2001). The DIMedS project can be characterized as an ambitious bottom-up project that did not manage to generate the necessary global mechanisms that would have allowed it to survive in its initial form (i.e. support education of surgeons). However, the activities in the project succeeded in creating something else: valuable competence and a control room that was well equipped for transmissions of a diversity of events and procedures. 4 Case # 2: Global information infrastructure development in a ship classification company (MCC) The second case describes a relatively successful development and use of a corporatewide information infrastructure (dubbed the Global-Scale Information System - GSIS) in a global ship classification company. This case has previously been described and analyzed from different theoretical perspectives (e.g. globalization and work transformation) by one of the authors (e.g. Braa and Rolland, 2000; Rolland, 2000; Rolland, 2003; Rolland and Monteiro, 2002). In this paper, however, we will mainly look at the role of power in the dynamics of developing and using the GSIS infrastructure, and how this is intrinsically linked to how the GSIS integrates previously separate activities and technologies, the strategic role of the project, and how it was embedded in situated practices of use.

12 4.1 The institutional and political context of the GSIS project As one of the largest ship classification companies in the world, MCC is a truly international organization employing 5500 engineers in over 300 offices in more than 100 countries worldwide. Founded in 1864, MCC is well known for providing technical inspections and risk assessments of high quality in a wide range of different industries. Ship classification, involving inspections of ships in operation, is one of MCC s primary business areas. Classification companies are part of an extensive network of institutions, international treaties and rules, established over the past 130 years, regulating the maritime industry in order to minimize loss of life and damage to environment and vessels. In the maritime industry, classification companies play a pivotal role in regulating and distributing risks among the different actors. More accurately, classification companies apply rules and expert knowledge in assessing the conditions of the ships, and hence act as a kind of brokering agency between the shipyards, the owners, the manufacturers, insurance companies, national and international regulating institutions. The stability with the historical role and the current structures of classification companies are rapidly changing. As shipyards and manufacturers are becoming grouped in larger units, and increasingly global due to recent mergers and acquisitions, classification companies and the classification system are put under pressure. As a consequence, the differences between the various classification companies are diminishing, and hence their core products become increasingly alike. This creates a pressure on classification companies to reduce the costs of their activities without compromising on quality and safety. In MCC, this renewed focus on safety and quality has been directly linked to use of information technologies in the sense that a company-wide information infrastructure is envisioned to enhance productivity as well as provide more knowledge to local surveyors that would improve the quality of the survey process. The primary job of MCC is to carry out annual, intermediate and renewal surveys (i.e. inspections) of ships in operation according to a complex set of rules and instructions set by MCC themselves, insurance companies, national and international regulating institutions (e.g. IACS and IMO). During a survey, a MCC surveyor conducts a detailed technical inspection of a ship, and if the focal ship is found to be in satisfactory condition and in accordance with the rules, the ship s certificate is endorsed or a new certificate is issued. Otherwise, a Condition of Class (CC) is issued, and if not fixed within a few days the ship s certificate can be withdrawn. In this survey process two important documents are made: i) A quick report contains a brief summary of what has been conducted and the results of the survey. This report is handed over to the crew after the inspection has been carried out. ii) A survey report is an extensive document which explains the survey in more detail and in which the surveyor writes his personal

13 comments and recommendations. This report is then sent to the ship owner and to the HQ. Until recently, the most important component of the infrastructure for supporting this survey process has been a centralized database system running on an IBM mainframe computer at HQ. The technical ship database (TSD) was originally designed in 1972 for keeping an updated file on each ship classified by MCC, and was extensively modified and extended throughout its 28 years of existence. The TSD was maintained by a small group of engineers at HQ, whose work was to extract relevant information from incoming survey reports written by surveyors in local offices. However, the TSD was not directly accessible to surveyors in local offices. During the 1980s technical information on ships was distributed on microfilm once every month, but later in the 1990s an application made it possible for downloading parts of the TSD on a local computer. However, many of the smaller offices could not afford the communication costs involved, and offices located in less developed countries without proper infrastructure could not access the database even if they could afford it. On the other hand, it is considered crucial by surveyors to have updated information on ships before conducting the survey, since in many cases the rules that apply depend on the ship type, tonnage, length, etc. Thus, in order to get hold of detailed information prior to conducting a survey, surveyors in local offices often had to contact a colleague at HQ to find the ship s paper file in the archives and fax them over. As a consequence, various reports and files were also stored locally in paper-based archives. In addition, during the 1990s in some local offices different IT-based systems were introduced in order to store ship information on a digital format. 4.2 Enrolling strategic allies The idea of a common globally integrated information infrastructure to transform and streamline the work of surveyors grew out of an increasing frustration of the fragmented and partly uncontrollable information infrastructure of MCC. At the same time, various research projects had focused on the development of generic and integrated IT solutions for use in the maritime industry and top management had initiated a project aiming at increasing global control of the survey process by centralizing decision-making in the company. In addition, in the late 1980s the CEO proclamation that knowledge about IT was as important as knowledge about ships and classification strongly legitimized and boosted research projects and grass-root initiatives to related all their activities to utilization of advanced information and communication technologies. An important point from the business managers point of view was to decentralize the activities of the HQ or to get rid of the expensive Norwegians as one manager put it, and in addition to make local surveyor work more efficient and uniform across

14 different offices. However in order to not loose control and to ensure uniformity, how a survey should be carried out, what information should be reported, in what sequence information should be reported, and how information should be structured in the final survey report were strongly inscribed into the vision of an integrated infrastructure. This also aligned perfectly well with MCC traditional focus on uniformity and standardization as main features of providing services of high quality (cf. Leidner, 1993). Besides enabling the local offices to access updated information on ships in a globally integrated model, the new information infrastructure was also planned to be a replacement for the old TSD database running on an old and expensive IBM mainframe computer. As time had gone by, after extensive modifications, ad hoc adjustments, and over 20 years of operation, TSD had become increasingly complex and expensive to change. Furthermore, a replacement was also urgently needed as the new millennium was approaching and IBM could not guarantee TSD being Y2K compliant. This emerging situation and the relevant actors interests the gave an opportunity for an enthusiastic group of engineers at the HQ to translate the various interests of relevant actors (e.g. the CEO, managers of important regions of MCC, local engineers at HQ, and surveyors) into a vision relating and promising to solve most of MCC problems through the development of a common integrated IT infrastructure. As a result, a technically-driven in-house software development project was established in 1994 that immediately combined the previous research projects emphasis on integrated models for ships (i.e. product models), their own interests in component-based software engineering, and the CEO and top-management s planned implementation of Microsoft technologies and solutions. The project was initially estimated to 8 million EURO and worldwide rollout of GSIS was expected to begin in early The ambitions and their adjustments Having translated the various actors problems and aims into an organizing vision of a globally integrated information infrastructure (see figure 1), the actual realization of the GSIS infrastructure implied highly unforeseen but nevertheless necessary negotiations, de-tours, and changing alliances. First, until1995 the research department and the IT department were both keen on implementing a UNIX-based operating system as the corporate-wide standard for MCC. This was very unfortunate as it threatened the very existence of the Microsoftbased GSIS project. The research department wanted a UNIX-based operating system since all their calculating tools were developed on this platform and the IT department also had considerable skills and knowledge on UNIX and equally little experience with Microsoft s Windows NT. This emerged as a manifest conflict as the GSIS project needed a 32 bit platform required for implementing the component-based solution,

15 which was at the heart of the design idea. To overcome this obstacle, the GSIS project, through personal contacts and persuasions, strengthened their network of alliances and enrolled external experts and companies to support their decision of Windows NT as the only viable alternative for a corporate-wide standard. Evidently, the CEO settled the matter by imposing an IT strategy strictly standardizing on Microsoft technologies. In 1997, due to the growing dissatisfaction in the GSIS steering committee and topmanagement because of numerous delays caused by underestimating technical complexity and the ongoing negotiations described above, the IT department sized the opportunity to launch their own application for supporting the work of surveyors. This application was based on already much used technologies like Microsoft Word and a newly implemented system. This application primarily developed using a simple script-language and re-configuring existing Word templates, it became an irresistible solution for surveyors to electronically submit their reports to the HQ. As a result, the application diffused like wildfire throughout the organization and thus challenged the current hegemony of the GSIS project. This situation lead to a resource demanding battle of systems there one the one side, the GSIS team argued that their solution was much more complete and technically sound compared to the more technically simple and pasted-up solution by the IT department. However, the actual implementation of GSIS relied on implementation of a common Windows NT infrastructure the very same infrastructure needed for the IT department s application. Subsequently, the IT department s application, although unintentionally, speeded-up the implementation of the Windows NT platform in local offices which in turn made the implementation of the GSIS infrastructure easier. Thus, in late 1998 MCC had long implemented the necessary underlying Windows NT infrastructure for the GSIS, and top-management and local offices had for more than a year been waiting for the highly acclaimed tool to arrive. On the other hand, the GSIS implementation project refused to implement the GSIS as they argued the current version of the GSIS did not yet have the needed functionality. Some of the members of the GSIS software project found this unacceptable, and argued that the users did not need a full-blown version of the GSIS in the first round. Through this translation the GSIS project managed to persuade the top-management to re-organize the GSIS project, so that the implementation project lost their mandate in the process. Consequently, in December 1998 the GSIS was introduced in offices that were perceived as allies with the GSIS project team. However, this solution also implied that other offices argued that the first version of the GSIS as one regional manager put it was totally stripped for functionality and at best useless maybe even hazardous for the quality of the survey process, and hence refused to implement this version of GSIS. The background for this rejection however, was more that this particular region due to their commitment to reporting to some of the regions largest shipping companies did not align well with the standardized way of reporting as inscribed in the

16 first version of GSIS. Thus, implementation of the first version of GSIS would for this region imply additional costs in adjusting the GSIS produced reports to local needs and thereby risking the local offices surplus. This re-alignment of the GSIS project to introduce the new infrastructure in some preferred offices, albeit the suspicions of insufficient functionality, turned out to be relatively successful in establishing an initial installed base of users and to test the current design of the GSIS. Over time, however, with increasing number of offices and users connected to the corporate-wide information infrastructure, the complexity increased and it became clear that the GSIS not first and foremost lacked functionality, but that the existing functionality was too rigid in that it imposed a too procedure-like work process to be followed by the surveyors. For example, the standardized checklists included in the system required to be filled-in in order to be able to submit the final report, and the level of detail in these lists also forced surveyors to look at the entire ship. In addition, a range of different best practices and global policies was established in order to ensure uniform use of the infrastructure. For example, there was a policy that required reports to be submitted within 24 hours, which limited the surveyors time to work around the rather imposed structure of the GSIS. Moreover, the GSIS infrastructure was designed in such a way that it also required a certain configuration of the underlying local infrastructure. Together these elements form a network which was disciplining surveyors use of the infrastructure. On the other hand, surveyors themselves applied different strategies to escape the inscribed pattern of use. For example, surveyors developed local variants of the checklists and templates for survey reports, and stored parts of the reports locally. Later, during 2000, the GSIS project opened for alternative and more flexible ways of reporting and thereby translating the interest of the surveyors and local managers. This turned out also to have side effects in terms of creating a need for local storage of digital survey reports, which then re-introduced fragmentation of information resources. In order not to loose the support for further development of the now highly strategic GSIS infrastructure, the GSIS project for the first time, slightly changed their overall vision and introduced an alternative for storing reports in local document databases (Rolland, forthcoming). Thus, in order to legitimize this move, the chief engineer of the GSIS project actively promoted the GSIS as a knowledge management system in contrast to a production system which it had been called until this fragmentation occurred. 4.4 Results Like most large-scale IT and infrastructure projects, the development of GSIS was fraught with all kinds of surprises, uncertainties and side effects (cf. Ciborra et al., 2000), but perhaps unlike many other comparable systems the GSIS was successfully put into use by over 1500 users in in over 150 of MCC s offices world-

17 wide. In 2000 MCC delivered their best financial result ever, and at the time of writing, GSIS is considered a huge success by MCC s managers claiming that the system has contributed significantly to MCC s financial success. Despite this success, however, business managers at HQ still argue that the GSIS infrastructure did not turn out as anticipated and that there is a need for developing a better underlying database model that integrates even more aspects of the survey process and avoids the current fragmentation of information stored in different databases and document achieves in local offices. Most surveyors also argue that the existing infrastructure does not always align with their practices and ways of dealing with complex survey jobs and customers who slightly deviate from the average. But, surveyors also claim that the current version (version 4) of the GSIS infrastructure is much better than the previous paperbased infrastructure. 5 Interpreting the cases 5.1 The practice of development: performing power and disciplining users Based on the case studies, one can perceive the issue of power as performed and changing and not a zero-sum game there one actor gains power at the expense of another. In this cases there are constantly changing coalitions and the different actors attempt to enroll other actors both human and nonhuman in order to strengthen their networks to support particular regimes of truth. Furthermore, common to the two cases described above is the need for ongoing work in order to keep the different actors aligned and the project on track. In the telemedicine case the main work consisted of generating use of the broadband link. This required approaching potential allies and enrolling them through offering a fully supported trial of the new technology. The prospective users had to be convinced that this was a viable technology that the hospital as a whole should choose, and the notion of high-end technology offering superior image quality was significant in this process. For the GSIS project in the classification company we see that this work of alignment took a slightly different form. Here there were a need to fight real or potential competitors and rivals in order for the project to survive and attain its goals. In addition to becoming the organizational actors chosen strategy, the project had to manage to discipline the different users. The choices and the actual design of the information infrastructure aimed at disciplining its use in both implicit and distributed ways. However significant this political work were in the two cases, it is interesting to note that such political work is generally:

18 Often not considered up front. The project is conceived as a rational project where technological considerations are important. Disputes around power issues during a project are seen as irrational, counter-productive, and negative. Also post hoc it becomes rationalized in the sense that the enactment of power is perceived as rational or the only alternative at the specific point in time; Yet the development of the infrastructures in question was dependent upon the enactment of power. The choices made in relation to actual rivals and allies strongly shaped the technical and non-technical characteristics of the project. The power did not reside with one group or actor only; neither does it seem to be a zero-sum game where one actor gains power at the expense of another. We should rather think of power as performed and changing. 5.2 Big is beautiful and dangerous Large and ambitious visions and project plans serves to draw the attention of potential allies, and to convince them of one the one hand, the potency of the initial actor, and on the other hand of potentially large payoffs if they participate. Thus the grand visions are of crucial initial importance in mobilizing partners and organizing the projects (Ellingsen and Monteiro, 2001). The two cases illuminate how large and ambitious project plans tend to be an integral part of integration technologies and information infrastructure projects. These technologies have an inherent logic of scale, due to the promise of benefits and payoffs when everything is integrated or connected. Our main point here is that such grand plans have as a consequence that the political work in the project takes on an unprecedented significance. Such projects by necessity involve the need to juggle a variety of changing alliances, to give and take, and to exercise power more or less openly. During this process the actors will make choices that will have more or less binding implications for future actions. The irreversibility of information infrastructures due to inscriptions in the technology has been studied (Monteiro and Hanseth, 1996). Based on our studies of information infrastructure development in telemedicine and in MCC, we argue that such irreversibility may also emerge merely from the political investment behind the stability, from the amount of political work that lies behind agreement. As seen in the MCC case, this has profound implications as the project s visions and strategies for implementing organizational change is not notably modified along the way even though the GSIS infrastructure and its development process has numerous unanticipated and surprising outcomes. This indicates that the size of the project seems to reinforce the vision which at first is helpful in mobilizing support for a specific solution, but in turn, becomes a hindrance