Collective Development in Open-Source Communities: An Activity Theoretical Perspective on Successful Online Collaboration Andrea Hemetsberger and Christian Reinhardt
Abstract Online collaboration is often organized without strong predetermined rules or central authority, which is why coordination and ways of organizing cooperation become crucial elements of collaboration. This article investigates how online projects can overcome problems of dispersed work, solve inherent contradictions and utilize tensions in the activity system to develop collaborative artefacts and practices. Empirical evidence is based on a detailed observation of a successful open-source project — the K Desktop Christian Reinhardt University of Innsbruck Environment (KDE). Our findings show that successful collaboration is based on coattailing systems. Coat-tailing means to inextricably bind together individual action and School of collective activity through careful design of complexes of technological, mental and culManagement, Austria tural artefacts. Andrea Hemetsberger University of Innsbruck School of Management, Austria
Keywords: activity theory, co-configurative work, online collaboration, open-source
The Internet has opened up new space for boundary-spanning collaboration and has generated invaluable technological improvements for the coordination of online projects. Organizations have embraced these promising opportunities for global collaboration with external partners, and for user integration in new product development. Victor and Boynton (1998) have coined the term ‘coconfigurative work’ for these new, technology-enabled forms of 21st-century, post-bureaucratic, networked conglomerates, wherein organizations, users and other external actors share resources, and collaborate. Organization theory has just begun to explore the potential, the idiosyncrasies and the drawbacks of these new forms of online collaboration. In such online work groups, work is often organized without strong predetermined rules or central authority, which is why coordination and ways of organizing cooperation become crucial elements of collaboration. This article introduces the notion of ‘coat-tailing’ — a term used to denote the parallel pursuit of individual and collective objectives — as a successful mechanism for online coordination and cooperation in co-configurative (Engeström 2004) online projects. Our central argument is that online cooperation is not just a matter of task coordination but rather a question of overcoming tensions that derive from the alignment of strategic activity and individual action within a highly dispersed group. Coat-tailing means to inextricably bind together individual action and collective activity through careful design of complexes of technological, mental and cultural artefacts.
988
Organization Studies 30(09)
We develop our argument based on a case study of an open-source project — the K Desktop Environment (KDE) — that employs a complex activity system for coordination and cooperation. Free and open-source software (F/OSS) projects — such as KDE and Linux, for instance — have been portrayed as the most global and successful examples of user integration and online collaboration (Lee and Cole 2003; von Hippel and von Krogh 2003; von Krogh et al. 2003; von Krogh and von Hippel 2006). In thousands of F/OSS projects, programmers collaborate online on a large-scale basis and produce software with unprecedented speed (Cubranic and Booth 1999). Despite the difficulty of mediated communication and work in a highly dispersed group, F/OSS projects master complex group tasks without regular face-to-face contact (Barthelmess and Anderson 2002). Programmers with various levels of expertise and users develop new software by sharing code and collaborating in a decentralized, selfdirected and highly interactive process (Kogut and Metiu 2001; Raymond 1999). Despite the benefits to be gained through the integration of users in product development, such volatile work organizations also face potential drawbacks through differences in cultural backgrounds and interests, and increasing complexity (Bechky 2003; Kellog et al. 2006), which cause conflicts, tensions and coordination problems (Orlikowski 2002; Metiu 2006). Successful F/OSS projects can teach us how to cope with these potential pitfalls. This article is intended to shed light on how co-configurative projects can overcome problems of dispersed work, solve inherent contradictions and utilize tensions in the activity system to develop collaborative artefacts and practices. To this end, we draw on important theories of ‘collective activity’ in the context of human-computer interaction, and introduce cultural-historical activity theory as our theoretical framework. Based on the review of this literature and F/OSSrelated research, we analyze the complex activity system of the KDE project in depth, and discuss the ways in which our findings contribute to activity theory and online collaboration.
Human–Computer Interaction and Computer-Mediated Collaboration Studies in human–computer interaction and computer-mediated collaboration in general show increasing attention to what people do in practice. They are based on the idea that both human beings and technology can only be understood within a social context. Theories of distributed cognition, situated action, activity theory and actor-network theory are prominent approaches for studying activity and action in context, be it situational, social, or in a network. The theory of distributed cognition moves cognitive psychology to the collective level and adopts a system perspective on goal-oriented, coordinated action (Hutchins 1993, 1995). It has embraced the notion of embodied cognition, arguing that material and social structures are elements of cognitive systems as well. Humans are adaptive systems, continually producing a rich world of cultural structures (Nardi 1996). Understanding this system means investigating the coordination among the agents as well as the instruments they use in order to reach their goal (Nardi 1996; Faraj and Sproull 2000). Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
989
In criticizing purely cognitive approaches, situated action models (Lave and Wenger 1991; Suchman 1987; Hutchins 1995) put particular emphasis on the flux of ongoing activity and the situational, sometimes improvisatory character of human action in particular environments. The emphasis is not on functional aspects or representational states, but on the relationship between the individual and their environment, and how it emerges. This view of the emergent and improvisatory character of human action is particularly powerful for investigating new, less structured forms of networked organizations. While situated action approaches are well suited for in-depth investigation of small socio-technical systems and micro-processes of situated knowing, they leave out broader perspectives of complex activity systems and how they account for successful collaboration (Engeström 1993). Situated action theories emphasize what people are doing in practice, but tend to lose focus of what people want to achieve as a collective entity. Because of its focus on technology, actor-network theory (ANT) has become a further prominent approach for research into computer-mediated collaboration. It proposes a dynamic view of interaction between society and technology (Latour 1996). The idea of the flexible, ongoing action in ANT is particularly appealing with regard to human agency in the Internet, where individuals immerse themselves in another world (Turkle 1995) and the difference between human and technological actors becomes blurred. Yet its symmetrical view of human and non-human action has also been criticized (Tuomi 2001); its emphasis on structures and social practice being inscribed in technology de-emphasizes the extensive human agency on technology itself that happens in online communities (Orlikowski 2000). Structuralist approaches have rather emphasized the situated use of technologies. When users interact with technologies they also decide how to interact with technology (Orlikowski 2000). Activity theory takes a similar approach in that it defines activity as the central unit of analysis (Vygotsky 1978). Cultural-historical activity theory argues that social practice should be understood as tool-mediated activity (Leontiev 1978, 1981; Cole 1999). This idea of mediation via tools is central to activity theory (Kaptelinin 1996). At the primary level, tools include physical tools that mediate people’s thoughts and behaviour. Conversely, people’s thoughts also shape technological artefacts and their usage. Vygotsky (1981) argued that mental tools are artificial formations and social by nature. They comprise language, mnemotic techniques, schemes, maps, drawings, signs, and other mental artefacts. More recent conceptualizations of activity theory have moved to a collective, artefact-mediated and object-oriented definition of the activity system (Engeström et al. 1997, 1999; Nardi 1996; Cole et al. 1997). Engeström introduced the community as the collective that is interested in an object, rules which mediate the relationship between a community and the subject of an activity, and division of labour as the way the community is related to the object of the activity. Objects are not to be confused with goals. While goals are intentional and drive action, the object (the Gegenstand of the activity) is the purposeful intended target or idea that includes the collective motive for the activity. Activity theory suggests three interrelated levels of interaction — coordination, cooperation and co-construction (Engeström et al. 1997; Bardram 1998). Coordination ensures that what people are doing independent of each other results Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
990
Organization Studies 30(09)
in the achievement of the common task (Engeström 1987). Cooperation concerns how coordination is achieved, and involves the social interaction of group members when doing things together. Co-construction corresponds to the re-elaboration or re-construction of work practices and demands reflective communication on a meta-level (Engeström 1987). At this level, work itself is the subject of contemplation (Barthelmess and Anderson 2002). By distinguishing these levels, activity theory draws particular attention to these boundaries and how they are managed in work groups. In order to manage these boundaries, artefacts need to be established that serve as anchors between different levels. Anchoring is an important concept and focuses on the interconnectivity between mediational artefacts in order to solve inherent contradictions of an activity system (Engeström 2006a,b). Activity systems need to link everyday activities upwards to the activity as a whole. Accordingly, the overall objective needs to be translated — anchored downwards — to the level of the concrete task (Engeström 2006a,b). Activity theory is an approach that explicitly focuses on the dialectical aspects of the activity system. Research is directed towards contradictions within the system, and discourse as an important catalyst for change and the co-construction of meaning (Wells 2002; Engeström and Blackler 2005). Through collective reflection, cooperative processes become visible and give rise to awareness for the need for improvement. Individuals develop a consciousness regarding their collective actions, and thus are able to initiate change and alter collaboration. By focusing on contradictions in existing activity systems, activity theory is particularly well-suited to identifying how complex and highly dispersed activity systems cope with these contradictions and how they use them for improvement.
F/OSS Projects as Co-configurative Activity Systems In an attempt to address new forms of collaboration, Engeström (2004) integrated the concept of co-configuration work into activity theory. Co-configuration is a participatory model that is not confined to collaboration between professionals, and integrates users as active subjects. Users are active in the shaping and reshaping of products and eventually become experts themselves. F/OSS projects are prototypical examples of co-configurative work, including professionals, experts and users. Collaboration among people with such varying expertise necessitates a dynamic, dialogic relationship between multiple actors; it is a relationship characterized by collaborative and discursive construction of tasks (Engeström 2004). Such groups are radically different from conventional teams or communities of practice (Lave and Wenger 1991; Nardi et al. 2000) in that membership at the periphery is fluid. In co-configuration work, participants are required to recognize and engage with different goals of action and different expertise distributed across group members. Work in F/OSS projects is voluntary; task assignment and decisions cannot be enforced (Demil and Lecocq 2006). Hence, conflicting goals of different actors could impede the pursuit of the activity and the achievement of the object. As a consequence, it is easy to lose sight of the overall objective in such complex collaborative online projects (Blackler et al. 2000). Apart from the extraordinary ideological and motivational grounding of F/OSS projects Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
991
(Raymond 1999; Hertel et al. 2003; Zeitlyn 2003), researchers report of distinct behavioural patterns, rules, norms and use of technology, which help coordinate different goals and tasks (Lee and Cole 2003; von Krogh et al. 2003; Hemetsberger and Reinhardt 2006). Researchers describe F/OSS projects as evolutionary, open, flexible, and thus creative interaction systems (Tuomi 2001; Lanzara and Morner 2003; Kuk 2006). Actor-network based research has portrayed F/OSS projects as ‘technologybased regimes’ (Lanzara and Morner 2005: 89). Lanzara and Morner (2005) argue that ‘technology can replace formal organizational rules and structures in the coordination [....] of complex activity systems, by inscribing large components of collective human agency’ (Lanzara and Morner 2005: 67). The authors maintain that the artefacts programmers use can be seen as attractors that organize activities in certain development areas and channel them towards certain development paths. Development is organized by technological artefacts, such as concurrent versioning systems of the code base and mailing lists, but also by licensing agreements. Lanzara and Morner (2005) and Hemetsberger and Reinhardt (2006) emphasize the coordinating character of source code itself. They depict code as materialized form of knowledge and a playground for variation and innovation. Mailing lists allow for communicative coordination in particular ways; licensing agreements inscribe legal and contractual rules. Innovation evolves through variation, selection and stabilization (Lanzara and Morner 2005). Lee and Cole (2003) particularly focus on the role of criticism and peer reviewing as mechanisms through which innovations are continuously created, selected and retained in F/OSS projects. F/OSS researchers have reported concordantly that openness of source code and open communication in mailing lists is not only the ideology but also the rule, and is considered key to successful coordination. Parallel and overlapping activities are possible without losing sight of the current version of their work (Yamauchi et al. 2000; Kuk 2006). Furthermore, it enables incoming members to re-experience how code was built, as well as how contributors collaborate in terms of rules and division of labour (Tuomi 2001; Lanzara and Morner 2003, Hemetsberger and Reinhardt 2006). Tasks are of modular character and are self-selected by contributors according to their expertise and interests (Lanzara and Morner 2003; Lee and Cole 2003). Although openness and modularity decreases complexity for the individual, it increases the complexity of the overall activity. Most researchers agree that parallel software development, peer review, and user involvement combined with ‘openness’ are the most important ingredients of F/OSS collaboration (Feller and Fitzgerald 2001). What has been left open is how F/OSS projects cope with the tensions that the pursuit of individual goals and collective activity entail. Our research aims to explore this question in order to gain important insights for the design of co-configurative work in highly dispersed groups.
Methodology This case study is part of a bigger F/OSS research programme. We moved from a first broad research question — how do members in F/OSS projects collaborate and integrate new members? — to a more detailed account of F/OSS projects as Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
992
Organization Studies 30(09)
complex, techno-cultural activity systems. Our research proceeded in several phases. First, we selected a case that provided deep insight into the collaboration process. Secondly, data had to be gathered over a sufficient period of time in order to achieve theoretical saturation (Goulding 2002). Thirdly, and in parallel with data generation, we were constantly involved in writing memos, coding, discussing emergent categories and interpretations, and analyzing data. Prior to the selection of an appropriate F/OSS community for investigation, we defined three requirements. We looked for a successfully established project that had developed appropriate technology and communication structures. In accordance with Crowston et al. (2004), we used (1) the time of existence, (2) the number of members, and (3) the rate of innovation and diffusion, as indicators. After an extensive exploration of other open-source projects and careful consideration, we ultimately selected the K Desktop Environment (KDE) project for our research purpose. KDE is a desktop environment for UNIX workstations, similar to those found under MacOS or Microsoft Windows. The lack of a user-friendly contemporary desktop environment has prevented UNIX from becoming a popular choice among typical computer users in homes and offices. In 1996, project founder Matthias Ettrich and a few programmers started a joint enterprise and produced the first beta version only one year later. Today, KDE is available in its fourth generation. KDE is one of the largest F/OSS projects in the world. More than 1000 developers around the globe are working and communicating via the Internet. To date, their collaboration has resulted in over four million lines of code. KDE is shipped with almost every Linux distribution. Research Method and Analysis
We chose a grounded theory approach (Goulding 2002; Charmaz 2006) adding non-participatory elements from Kozinets’s (2002) netnography. After having chosen KDE, we asked the KDE project leader for permission to observe the community and their communication. We closely monitored the project community for a four-month period in order to gain a deep understanding of what they were doing. We observed the community regularly, two to three days a week, looking at Websites, tracking chats and scanning the mailing lists. In addition, the research team familiarized itself with the KDE desktop environment in order to understand the ‘work philosophy’ of its creators. We also included external open-source affiliates in our discussions and attended F/OSS conferences in order to observe their culture and grasp their technical jargon. Making our theoretical perspective explicit and focusing on activity theoretical concepts, such as coordination, cooperation and co-constructive events, helped us concentrate on collective core activities and contradictions with individual actions that occur in F/OSS collaboration. We carefully avoided using preconceptions, for instance theoretical explanations, in analyzing the data (Charmaz 2006). After a first period of intense observation, the particular importance of mailing lists for the pursuit of the core activities in the KDE project became evident. More specifically, we focused on general development lists which have the highest impact, and focus on the core activities of KDE. We applied the procedure of open, axial and selective coding as described by Goulding (2002). After initial coding we entered a focused phase of analysis. We Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
993
sampled relevant threads for further analysis and interpretation of the emergent categories. As the subjects of the threads are mostly task-specific and thus not selfexplanatory, we selected relevant threads via key word search, and by scanning all threads. We analyzed 30 threads in depth, including 278 contributions from 143 authors to the general developers mailing list, and 136 contributions from 46 authors to the core developers mailing list posted in February and March 2003. The core developers mailing list is used for discussion about the development of the main developer tools and strategic activities, whereas the developer list is used for general discussions regarding KDE source code development. We selected a time span which represents a period of ‘normal’ activity in the community. Periods before a major release (major releases are releases with a change in the second digit as, for instance, the release of KDE 3.1) are mainly concerned with bug fixing. Therefore, we chose a period after the release of KDE 3.1, but including the next KDE 3.1.1 service release. For the purpose of studying co-construction, we engaged in theoretical sampling of instances of contradictions that led to changes of work tools. We additionally analyzed 10 corresponding threads spanning from August 2004 to October 2004, including 360 postings on the core developers mailing list concerning KDE’s switch of version control systems (the most important tool for codebase management). We sent four short email questionnaires to four core members, including two developers who initiated the change, and two who are responsible for the version control system. Their answers informed our data and interpretation with motives for the initiation of this major change, with technical details, timing issues and explanation of rules. Once we had achieved theoretical saturation (Goulding 2002), we integrated the findings into a coherent theory. To test our theory, we asked for feedback from community members in a standard procedure of ethnographic research methodology called ‘member check’ (Kozinets 2002). Findings The KDE Activity System
From an activity theory perspective, the KDE project unfolds as an activity system with a clear object as formulated in their publicly announced manifesto (see: http://www.kde.org/whatiskde/kdemanifesto.php). The manifesto is not only an expression of their ideology and will, but also important for the self-selection of volunteer contributors because it has the potential to reduce the number of different motivations in the group. By making the object of their work explicit, F/OSS communities, although open to the public, restrict access and delineate their activity system. In general, the most distinctive characteristic of F/OSS communities is that they are open to anyone willing and able to adhere to their objective. User integration is therefore a core activity and is strategically important. KDE’s main activity concerns the constant improvement and development of their product — the codebase, which mainly involves cooperation and coordination of individual contributions. Development, in turn, needs innovation in terms of excellent ideas and excellent developer work tools. Innovation is KDE’s third core activity on a collective level of activity. Table 1 outlines our main findings. Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Collective innovation Joint conceptualization of ideas Co-constructing the activity system
Collective development Concerted development Mailing lists as ‘help desks’
Consensus and volunteering Collective reflection and decision-making The power of the doer
Granting write access Code reuse Peer review and feedback
Omnipresence of the object of the activity Open (reading) access to communication and source code Voluntary assignment and selfselection of tasks
Cultural artefacts
Mailing list discourse and archives, IRC channels Rare, but strategically important face-toface meetings
Bug database, notification emails Version control system, modular structure Commits-list, mail filters, diff application Mailing list discourse and archives
Structured content: Blogs, Wiki, Faqs, Web site content, documentation Sequential content: Mailing lists and archives; online tutorials Bug database, wish lists
Technological artefacts
Activities and Artefacts in the F/OSS Activity System as Enablers of Collaboration
New member integration Cultural entrée Learning and adoption of programming style Continuous improvement
Activities
Table 1.
Mental experimentation Idea reviewing
Productive inquiry of code ‘Tech-talk’ as language artifact
Retrospective experience Direct experience Productive inquiry
Mental artefacts
Idea generation and shared understanding Democracy and meritocracy
Productive redundancy and recycling Open reflection and complexity reduction
Attracting new members and avoiding distraction Volunteering and task prioritization
Enabling
994 Organization Studies 30(09)
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
995
New Member Integration
The integration of expert users is at the core of co-configurative work and has obvious advantages for F/OSS communities. First, user feedback is integrated into innovative software solutions. Secondly, it increases opportunities by enlarging the workforce. However, for user integration to be successful, coconfigurative workgroups have to cope with some inherent contradictions. Users must be made aware of the opportunities to contribute. Yet, users have different expertise with regard to usage, and programming. Hence, co-configurative work needs educational facilities starting at different levels. While communities of practice apply an apprenticeship approach where mentor and apprentice work closely together, F/OSS project members have to rely on remote training and interactive help from more experienced programmers. However, attracting new members should not interfere with the primary task of programming. The goal of user integration, therefore, is to integrate aspirant members with different expertise, but to avoid distraction from pursuing core activities. Furthermore, F/OSS projects have to solve the tension between volunteer work and task prioritization. The KDE community tries to solve these contradictions through designing the cultural entrée, providing learning opportunities at different levels, and integrating users in continuous improvement. Cultural Entrée in an Open Community
Aspirant members’ first contact with the KDE community is usually established through the project’s online contents. These contents mediate contact with the community in a very specific way. A clear mission statement, software downloads, access to the source code and all mailing lists, and the developers’ site symbolize openness, free access, and provide an explicit invitation to participate. Via a ‘Getting involved’ link, users and aspirant developers are invited to visit http://techbase.kde.org/ — the site which is dedicated exclusively to present and future KDE developers. A conglomeration of detailed descriptions and rules allows the newcomer to become familiar with the community and its activities. For coordination, self-motivation is one of the most important principles of the KDE development model. Hence, this is also expected from anyone who wants to get involved with the community. The following brief excerpt taken from the user mailing list is an example of that mentality: Hello Everyone, I am totally new to KDevelop, please let me know what it is, when I saw it for the first time I found it as if it is for developing new applications for KDE... Please tell me how to start with KDevelop... if I want to develop some Applications like what we do with visual basic on Windows platform then what is the best (Let me know whether I can do something with QT Designer for Developing Applications to run on Linux...)……………. Thanks & Regards
The reply reads as follows: Maybe you want to wait for Visual Basic for Linux? Perhaps it is available in about 50 years. Since you are new to Linux I can give you the astonishing advice to read the
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
996
Organization Studies 30(09)
documents which come with your system or are available at http://www.kde.org/. I do not believe that someone here has the time to write the documentation especially for you.... ...You can believe me that I do not expect this mail to help you, but I could not resist. Sorry! (retrieved from: kde-user 2001–08–15)
This discourse tells two important stories. First, KDE is an open, but also a political-ideological culture in its basic tenets. Individuals who are thinking in terms of competitor work tools do not pass the cultural entrée. As the objective and typical working style of F/OSS projects is firmly based on its cultural values, cultural discourse is particularly present in communication with newcomers. Second, the above excerpt gives clear and explicit advice as to where newcomers should look for their first contact with the community. The harsh tone of the answer echoes the writer’s scorn for those who disregard the rules. Here the community seeks to avoid distraction. The technologies used are simple: publicly observable sites, FAQs (frequently asked questions) and discourse. Nevertheless, there are still many cultural tensions with aspirant members. In order to overcome the ‘newbie’ barrier, the community constantly creates new and simple tools for newcomers that are geared towards the proliferation of culture and rules for collaboration. Since different newcomers invariably ask the same questions, the information they need can be pre-structured. FAQs and the Wiki site are collaboratively created over time and structured from the perspective of the ‘information seeker’. Aspirant developers also benefit from discursive events, stored chronologically in mailing list archives; they are encouraged to observe common practice and discourse before they become a member. Presenting discourse in a sequential order enables internalization of the cultural norms of the group and their way of thinking. Open communication enables newcomers to re-experience community history — which brings aspirant members a step closer to contributing. Learning and Adoption of Programming Style
KDE members know that, from the first moment on, participation must be intrinsically rewarding in order to attract volunteer contributors. Similar to those activities Csikszentmihalyi (1990) portrayed as autotelic, the community designs tasks which can be mastered quickly and provide immediate feedback and gratification. The goal is to provide help for different levels of expertise and to offer different opportunities to participate. KDE supports the first steps of integration by providing tutorials on: http://techbase.kde.org/. In contrast to documentation, which only provides an abstract description of how things work, tutorials are much more oriented towards the activity of using work tools and coding itself. Once newcomers are familiar with the community’s technology and programming style, they engage in small, simple tasks according to their level of expertise. Rules for first contributions are briefly described in the ‘How to help KDE’ for potential contributors. The site includes important cultural cornerstones that foster individual motivation such as the freedom to choose and to do what is fun. F/OSS communities must work with the constant contradiction of having a voluntary workforce while at the same time getting things done that Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
997
have to be done. Therefore, the community must find ways to encourage participation in activities that contribute to the common objective. Task prioritization is partly fostered by discourse which conveys the message that contributors should bear the community’s overall objective in mind. Beyond that, the community sincerely appreciates engagement in tasks that contribute to the common objective and give the authors credit for this on public contributors’ lists. Despite the myriad of information and the openness of discourse, newcomers are continually asking where to start contributing. As a matter of fact, KDE has constantly struggled with giving the right advice to newcomers willing to contribute. This contradiction has triggered discourse and the introduction of dedicated mailing lists and websites (http://quality.kde.org/). Yet, quality.kde.org does not convey the intended meaning, as reflected in the following Blog entry. I think that the name ‘quality team’ was a historical mistake, though. While the intention is to get new contributors to work on the quality of KDE as a whole by dealing with polishing, small bugfixes, documentation &c., but the name is poorly chosen. (retrieved from a member’s Blog: 2006–05–08)
Language artefacts have to be chosen carefully in order to deploy a self-selection mechanism. Iterative fine-tuning of language artefacts is necessary to resolve these contradictions. Hence, the community has started to call small tasks for beginners, such as bug fixes, ‘Junior Jobs’. Initial contributions are further encouraged by propagating a pragmatic approach of doing. It fosters, of course, learning by doing, but also prevents newcomers from doing ‘invisible’ work. Hence one of the core tenets of the KDE philosophy reads as follows: When making a suggestion, change ‘we should..’ to ‘I will..’; grandiose plans are useless unless you are willing to put in the work. (retrieved from: http://www.kde.org/whatiskde/devmodel.php)
As the Internet is a space where individuals are only visible through contributing content, doing means visibility and thus, progress. This particular meritocratic approach is enabled by several work tools, including bug tracking technology. Continuous Improvement
The specific way F/OSS projects handle bug fixing is reputed to be one of the main advantages of their development model (Kuk 2006). The primary advantage of the bug tracking system is that the detection of bugs and bug fixing are split among individuals with different expertise. Having a large group of advanced users testing the software helps in finding problems quickly. This makes the system exceptionally well-suited to fostering self-selection of high priority tasks. This is also why the bug tracking system has not been changed but rather only slightly improved in its basic functionalities. The bug tracking system is the work tool that enables advanced users to report a bug or a missing feature in a way that enables developers to fix it. User voting for the most hated defect or the most desired feature further reflects the severity of the problem, which helps with the prioritization of tasks. Bug fixing is a delimited operation modularized from the overall codebase and thus a good opportunity for new developers to contribute and improve their Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
998
Organization Studies 30(09)
programming skills. New members, however, are not allowed to commit a bug fix — in F/OSS jargon a ‘patch’ — directly into the source code repository themselves. The maintainer reviews bug fixes before integrating them into the code or rejecting them. New members benefit from receiving comments on rejections, much like in academic reviewing. This results in the continual improvement not only of code, but also of developers’ skills. Reviewing and feedback are invaluable mechanisms which link individual action to the collective activity. > You might consider asking for SVN write access so you can commit these > fixes yourself.... Well, I just wanted to see how well am I doing before getting an account:). When I think I’m ready, I’ll to it. Perhaps, it is time:))). Thanks, I’ll look into it. (retrieved from: kde-devel 2006–04–13)
Developers who have reached an acceptable level of competence are invited to ask for write access to the source code repository so they can integrate patches to the source code. Being granted write access entails strong social symbolism. It engenders pride on the side of the ambitious developer, expressed here by the use of emoticons, as it marks the transition from a newcomer to an integrated member of the KDE project. Hence, it is an important cultural means to foster excellence. Collective Development
In their manifesto the KDE community commits itself to excellence. The outcome of their activity should be no less than the best desktop environment. Continuous improvement is one way the community tries to achieve this excellence. Peer reviewing and feedback contributes to this. KDE reuses peer reviewed code, because it prevents the community from reinventing the wheel. However, new ideas develop out of variation and continuous experimentation (Lanzara and Morner 2005). As a consequence, KDE set up technology and rules which enable both reuse and redundancy. This produces an enormous amount of code as well as discourse for collective reflection, which calls for the reduction of complexity. In the following, we describe how the community solves these contradictions. Concerted Development
While user integration increases the knowledge base and helps to evaluate the benefit of future software solutions, it also broadens the range of tasks and increases complexity. Task complexity is especially problematic when a huge number of developers are involved (Brooks 1995). Modularization breaks the whole source code down into manageable parts. Consequently, complexity for the individual is low, as members can work on modularized tasks which fit their expertise. Furthermore, specific modules can be used for several applications. The rule of reuse adds to this by forcing contributors to actively search for Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
999
modules that are peer-reviewed and can be incorporated in a further piece of work. ‘Use available tools rather than re-inventing existing ones!’ is one of the core tenets of the KDE philosophy. Modularity and reuse, combined, thus contribute to both an increase in variety and a reduction in complexity. As this dynamism is at the core of KDE’s collaborative work, corresponding rules are sometimes fervently defended, and violations provoke unfriendly replies: >> You saw that our library was not ready yet, so instead of helping >> finish it, you wrote a new one from scratch? > Yes, sorry, I wanted to have it working very fast. It took me a day to write > it and it’s far from being perfect. And now I’m here, don’t I? So you started a new one, which also is still imperfect. Classical lose-lose situation. (retrieved from: kde-core-devel 2002–05–08)
In F/OSS projects, several developers work on the same application, sometimes even on the same module. Concerted action through common artefacts (Bødker 1997; Bardram 1998) is therefore crucial. The KDE project uses the version control system SVN (Subversion). Continuous coordination systems (van der Hoek et al. 2004) avoid change collisions in the code base, therefore they are of utmost importance for the coordination and development of KDE’s codebase. Version control systems help people see at a glance what others are doing and hence improve visibility (van der Hoek et al. 2004). In order to make changes in source code visible, SVN provides a very simple tool with the ‘diff’ function. The use of different colours enables developers to quickly scan the changes made in the repository. As with almost every other technology used in the community, its complexity-reducing functionality lies in its simplicity. Such work tools not only coordinate work and encourage reuse, they also foster productive redundancy. Reuse is only feasible for good, reviewed work; new ideas develop out of continuous experimentation. Mailing Lists as ‘Help Desks’ and Sources of Inquiry
Mailing lists and IRC channels provide platforms to solve problems and collectively think about action. Developers reflect on their work and evaluate simple ideas for implementation. KDE is exploiting the fact that mailing lists open the opportunity for self-selection, whereby help is provided by those developers who think they are most likely to have a good answer. The communicative interactions that emerge can involve a variety of different contributors with different expertise. On the Internet it is not important to know who-knows-what (Faraj and Sproull 2000); it is sufficient to know where to post or to search for. As the addressee is the entire group of participants who are subscribed to a mailing list, helping behaviour in mailing lists is extensive. However, the fact that anybody can talk is both an advantage and a disadvantage. Written communication encourages reflection before someone hits the send button. Even so, core developers also admit that ‘social interaction on the mailing lists could be drastically improved’. Still, due to the fact that mailing lists are powerful ‘help desks’, they have never been changed in their basic functionality. Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
1000
Organization Studies 30(09)
However, the constant influx of new members and an increase in code and subprojects has led to several refinements of the system. Mailing lists split from time to time into new, topic-specific mailing lists in order to keep the amount of discourse on the lists within digestible limits. Furthermore, mailing list digests reduce the overwhelming number of mails received. Developers can also search mailing list archives. Archived discourse with experts constitutes an invaluable source for productive inquiry when developers look for solutions to their coding problems. This search comprises the use of theories, concepts, rules of thumb, combinations of these, and other mental reflections that mediate action. Content analysis revealed that developers primarily use source code, and error messages in particular, to express their problems. Another common way to express technical problems is to present solutions which have already been tried out, in order to make other developers understand the problem. Code is the common language of the audience and is in itself an invaluable mental artefact, because it reflects the mental models, abstractions and theories-in-mind of its creators. Collective Innovation
Together with the code repository, mailing lists build the project’s platform for technical help, but also for reflection, and idea generation. Collective reflection shapes goals and produces collective ideas that go beyond individual thinking. These collectively produced ideas are the source of innovation. Yet, collective reflection on new ideas demands shared understanding of the problem or issue involved, which is difficult to achieve online. The KDE community uses several mental artefacts for joint conceptualization of future ideas. Beyond ideas for software applications, KDE is also an innovator of development tools. We will report on one momentous development to show how contradictions in the activity system initiate co-construction. Joint Conceptualization of Ideas
Reflecting on problems and conceptualizing new ideas are important creative processes in the KDE community. Developers use the mailing lists to present their ideas, for instance for a new application or feature, and ask others to comment on them. Such initial messages generate lively and productive conversations, provoking comments, and comments on comments. Developers are aware of this functionality and actively exploit its potential. Content analysis showed that developers use various mechanisms to overcome the drawbacks of physical distance, for instance recapitulating an idea or elaborating the idea in more depth. They further support ideas, present different perspectives of the problem, point out flaws, insist on their views, disagree, and defend their own ideas in a constant process of idea generation and reviewing. However, for joint conceptualizations, KDE developers use other mental artefacts such as programming language (e.g. plain code; ‘what if, if then’ arguments), analogies, and future usage scenarios for collective reflection. One simple variant is to use analogies. Similar to what Bechky (2003) described, developers ‘de-contextualize’ their idea from the technical background and put
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
1001
their thoughts in a comprehensible, everyday context. Contrary to ‘tech-talk’, the language here changes drastically in order to achieve a common understanding of a problem. Analogies have their drawbacks, for instance when the analogy doesn’t hold. Yet, used as a mental artefact, analogies are enablers of collective reflection and evaluation of ideas. Another common method of supporting the presentation of an idea is to describe future outcomes or usage scenarios and collectively develop ideas in a way that describes a ‘virtual reality’, that is, plays with the future realization of the respective idea. That’s basically the idea. …….The idea is to have several standard KDE-wide wallets, and then allow applications to create their own as well. For instance, there could be a ‘network passwords’ wallet, a form data wallet, and a ‘local passwords’ wallet to start. This way the user doesn’t need to unlock his passwords to get form data.
Another KDE developer adds his own understanding and develops the idea further: This would be interesting, I would think that it would be a kde wallet manger which manages sub wallets. Any app which belongs to a wallet class would be shared. Apps with special concerns could then require a special class onto themselves. The manager would allow you to group classes under a single unlock function. So I could do email, fish and kdesu passwords all with one password. …….. (retrieved from the kde-devel 2003–04–30)
The use of language as a mental artefact enables developers to collectively create innovative ideas. In the course of the reflection, contributors engage in continuous mental experimentation with ideas. Discussants collectively work to circumscribe their ideas, thus trying to achieve shared understanding and setting goals for future action. These processes, of course, do not run smoothly. Analysis has shown that during discourse, ideas are constantly contested. In a way, ideas are peer-reviewed. Idea reviewing differs from the peer review of source code as described by Lanzara and Morner (2003) because of its emphasis on generating rather than selecting ideas. Using the reuse rule here would be detrimental for idea generation. Instead, for innovative and co-constructive tasks, the community has put a focus on doing, which provides the ultimate feedback — either it works, or it doesn’t! Co-constructing the Activity System
Work tools are changed or developed when there is tension, be it a technological or a social problem, or when new opportunities arise (Bardram 1998). Discussions on the object of work are extremely rare; however, the growth of the KDE code- and developers’ base has led to tensions that derive from the usage of work tools, such as the version control system. KDE originally decided to work with the ‘Concurrent Versions System’ (CVS). Their switch from CVS to SVN was one of the most far-reaching changes in recent years. Restrictions in reorganizing and renaming parts of the source code without losing a file’s history, for example, hindered developers from experimenting on more innovative software concepts in parallel. This contradiction triggered discussions among core developers. Strategic changes like these are initiated by the core developers team. Furthermore, decisions with such far-reaching consequences are first discussed face-to-face.
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
1002
Organization Studies 30(09)
A one-year discussion was held on the core developers mailing list in order to retrieve opinions from everyone concerned and achieve a shared understanding of the problems involved. The core developers list is the only list where participation is restricted to developers, but it is publicly accessible. This reduces the perils of an unmanageable amount of discourse and at the same time enables all interested members to gain a shared understanding of the ongoing issues. Analysis of related threads showed that the decision-making process follows a particular path. In a first step, all developers are invited to contribute to the discussion. Extensive information on SVN and other version control systems was collected, evaluated and discussed. Developers considered not only the impact of the new system on their future work but also the consequences of the switch itself. The rules of decision making are consensus and ‘he who does the work decides’. When mutual consent among KDE developers had become apparent, one of the core developers took over the lead and wrote a task list. This was followed by a call for active commitment. Task lists and commitments signal a switch towards action. After decision-making, further discussion is unwanted: > > Fortunately I can say, that subversion will open soon for performance and > > acceptance testing and if that passes, we will migrate our CVS as fast as > > possible (different thread!) > Why? What are the current shortcomings of CVS that cannot be overcome in > the future (moving modules?)? This is AFAIR no longer any part of discussion to prevent the move from CVS away to SVN. There’re pros and cons, as always..... (retrieved from: kde-core-devel 2005–02–14)
The community sets clear rules: democracy and meritocracy, not in a political sense, but as a rule as to how to do things. While democracy manifests itself through open discussions, meritocracy is the way in which those who commit themselves to work are empowered. Although seemingly contradictory, both entail important functions. As with the strategic decision, open discussions increase the knowledge-base and increase acceptance. After finalizing the decision, further discussions distract from implementation, thus meritocracy takes over. Discussion It has been argued that the world of work is going through some major transformations as global, decentralized, participatory, creative online organizations are taking shape (Engeström 2006b). Our study was intended to enrich the increasing body of research in the field of online collaboration and co-configurative work. It contributes to the existing literature in two important ways. First, we address and extend the recent assertion in co-configurative work research that the overall objective and the goals of activities of the many and diverse actors
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
1003
have to be ‘anchored’ in order to link coordination and cooperation in complex activity systems. Second, our research contributes to our knowledge of how to design online activity systems that are able to cope with the contradictions inherent in co-configurative work. Although limited to the case of F/OSS development, it has certainly helped broaden our knowledge about participatory forms of online collaboration, integrating users and various other external partners. A central insight derived from our findings is that online co-configurative work needs what we may call coattailing systems. Coat-tailing is used in different contexts, for instance it can be used to describe how to boost the success of action by connecting it to something which is already successful. In politics, coat-tailing describes the tendency for a popular political party leader to attract votes for other candidates of the same party. The success of cover versions (music that builds upon songs that have already been popular) has also been explained with the coat-tail effect. In a similar vein, coat-tailing work systems tie everyday actions to the overall activity of the group and thus couple the achievement of individual tasks with the success of the project as a whole. Due to the fluid boundaries of co-configurative groups, and their dispersed character, coat-tailing is of critical importance. As various actors are constantly contributing in parallel, activity systems have to be designed that weave the work of many into the web of activities. Coat-tailing systems enable ‘doing just one thing together’. However, coat-tailing is not an automatism. As depicted in Table 1, cultural, technological and mental artefacts need to be carefully combined to create a coat-tailing system. This in turn supports the main activities of an online group (far left column) by solving the major tensions within the activity system (far right column) that derive from differing individual and collective goals. Coat-tailing extends the concept of anchoring as described by Engeström (2004). Contrary to work environments where links between different levels of activities are established with particular single artefacts (Engeström 2006b), as for instance working schedules, in dispersed groups such boundary objects are prone to avoidance (Sapsed and Salter 2004). Dispersed groups, rather, develop what Kellog et al. (2006) have called a ‘trading zone’ by engaging in the practices of displaying, representation and assembly of work through the use of Internet technology. Our research supports this assertion. Coat-tailing means carefully aligning individual actions and collective activities by taking into account the motivations of the single human actor. Work tools and rules are designed to match individual expectations and to enable individual activity, where the collective activity is achieved by ‘piggybacking’ on the fulfilment of the individual task. Coat-tailing is a subtle effect as individuals are not necessarily aware of it. The bug reporting tool, for instance, invites newcomers to commit quick and easy self-selected contributions, while enabling prioritization of tasks at the same time. Creating awareness and visibility is of central importance in online work groups (van der Hoek et al. 2004; Kellog et al. 2006). Hence, coat-tailing systems need to be open, legible and visible. However, exactly because of these open and flexible structures, online collaboration often suffers from complexity (Orlikowski 2000). Therefore, coat-tailing systems are also restrictive and exclusive. First, as
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
1004
Organization Studies 30(09)
technology’s mediating capacity is strongly dependent on users’ understanding of the properties and functionalities of a technology (Orlikowski 2000), coattailing systems need appropriate technology for different users. Second, restrictions are also strongly conveyed in openly displayed discourse, content, rules and licensing. Openness thus contributes to access restrictions. Contradictory as it may sound, this is exactly the characteristic of coat-tailing systems: to embrace contradictions in order to solve them. This is why coat-tailing systems are not a stand-alone solution, nor should they be thought of as merely technological artefacts. Our findings corroborate structuration theorists’ contention that technology alone would fail to result in successful collaboration without the hidden and overt cultural rules (Orlikowski 2002). Technology and culture are closely interwoven to support the parallel pursuit of the long-term, collective activity and short-term, individual tasks. In order to resolve contradictions between individual action and the collective activity, combinations of artefacts are applied which flexibly adapt to different situations. Our findings support Engeström’s (2006a,b) proposition that co-configurative work tends to form integrated ‘toolkits’. As a coherent system they foster mental processes and enable action. Coat-tailing systems are geared towards making people do and think. While browsing through open content, newcomers, for instance, may discover interesting tasks and engage in productive inquiry. These insights also have some broader theoretical implications for the theories of situated action in that they add important insights into how apprentices are enabled to observe practice and to productive inquiry in mediated environments. Our findings add yet another component to online collaboration which is important for collaborative success in dispersed group work. Coat-tailing systems foster and allow for the pursuit of individual goals and the strategic objective. However, contrary to Engeström (2006a,b), who presents a hierarchical view of either upwards or downwards ‘anchoring’, we found that online coconfigurative systems are designed so as to enable both, and at the same time. Coat-tailing systems enable parallel processing and attainment of strategic objectives and operative goals. User integration, for instance, takes place simultaneously on the strategic level of cultural integration and on the operative level of immediate task involvement. Furthermore, coat-tailing systems are geared to cope with tensions that arise from the pursuit of both. The tensions as well as the solutions to the strategy–task conflict are inherent in coat-tailing systems. Yet, both are important; tensions are important triggers of change, while solutions enable coordinated work. Viewed from this perspective, coat-tailing systems are also important change agents. They enable co-constructive processes of permanent fine tuning of workflow technology, mental artefacts and discursive renegotiations of cultural artefacts. We conclude that, contrary to recent arguments of Lanzara and Morner (2005), technological artefacts do not become holders, but rather vehicles or, as we like to put it, enablers of human agency (Kuutti 1996; Orlikowski 2000). In a similar vein, we departed from the view that structures and sequences of human actions are inscribed in technology (Latour 1996). Although we share the view that technology plays a decisive role in online collaboration, we found that technology tends to follow the prescribed cultural rules and norms, rather than
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt: Collective Development in Open-Source Communities
1005
the other way round, particularly so in an environment where technology is developed by the collaborators themselves. By focusing attention on contradictions within an activity system as potential triggers of co-construction in work design, we also opted for a process perspective in order to overcome the shortcomings of a purely static view. It is not technologically sophisticated work designs but the ability of groups to co-constructively react to tensions in activity systems which contributes to sustainable online collaboration. Co-configurative work rests on informal relationships, liberal sharing of information, meritocracy and openness to external members. Literature has portrayed these particularities as paradoxical, as these organizations seem to act in a contradictory fashion (Kreiner and Schultz 1993). Yet, co-configurative activity systems are actually only contradictory when viewed from a coordinative perspective on the level of individual actions. Viewed from a cooperative perspective, the paradoxical nature of contradictory human actions vanishes. By embracing contradictions and weaving them into a coat-tailing system of technological, cultural and mental artefacts, online collaboration can transgress the limitations of the dispersed group work.
Note
References
We owe an enormous debt of gratitude to all the KDE community members for their openness, helpfulness and insightful comments regarding our work. May their spirit continue to enlighten their own lives and those of others. The authors are very grateful to the Senior Editor, Jacky Swan, and the three anonymous reviewers for their insightful and helpful comments.
Bardram, Jacob E. 1998 ‘Collaboration, coordination, and computer support: an activity theoretical approach to the design of computer supported cooperative work.’ PhD Thesis, Aarhus. Barthelmess, Paolo, and Kenneth M. Anderson 2002 ‘A view of software development environments based on activity theory’. Computer Supported Cooperative Work 11/1–2: 13–37. Bechky, Beth A. 2003 ‘Sharing meaning across occupational communities: The transformation of understanding on a production floor’. Organization Science 14/3: 312–330. Blackler, Frank, Norman Crump, and Seonaidh McDonald 2000 ‘Organization processes in complex activity networks’. Organization 7/2: 277–300. Bødker, Susanne 1997 ‘Computers in mediated human activity’. Mind, Culture, and Activity 4/3, 149–158.
Brooks, Fred P. 1995 The mythical man-month: Essays on software engineering. Reading: Addison-Wesley. Chaiklin, Seth, and Jean Lave (eds) 1993 Understanding practice: Perspectives on activity and context. Cambridge: Cambridge University Press. Charmaz, Kathy 2006 Constructing grounded theory: A practical guide through qualitative analysis. London: Sage. Cole, Michael 1999 ‘Cultural psychology: Some general principles and a concrete example’ in Perspectives on activity theory. Y. Engeström, R. Miettinen and R. L. Punamäki (eds), 87–106. Cambridge: Cambridge University Press. Cole, Michael, Yrjö Engeström, and Olga Vasquez (eds) 1997 Mind, culture, and activity: Seminal papers from the laboratory of comparative human cognition. Cambridge: Cambridge University Press.
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
1006
Organization Studies 30(09) Crowston, Kevin, Hala Annabi, James Howison, and Chengetai Masango 2004 ‘Towards a portfolio of FLOSS project success measures’. ICSE Open Source Workshop. Csikszentmihalyi, Mihalyi 1990 Flow: The psychology of optimal experience. New York: Harper Collins. Cubranic, Davor, and Kellogg S. Booth 1999 ‘Coordination in open-source software development’. Proceedings of the 7th IEEE International Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprises. Demil, Benoît, and Xavier Lecocq 2006 ‘Neither market nor hierarchy nor network: The emergence of bazaar governance’. Organization Studies 27/10: 1447–1466. Engeström, Yrjö 1987 Learning by expanding: An activity theoretical approach to development work research. Helsinki: Orienta Konsultit. Engeström, Yrjö 1993 ‘Developmental studies of work as a testbench of activity theory: The case of primary care medical practice’ in Understanding practice: Perspectives on activity and context. S. Chaiklin and J. Lave (eds), 64–103. Cambridge: Cambridge University Press. Engeström, Yrjö 1999 ‘Innovative learning in work teams: Analyzing cycles of knowledge creation in practice’ in Perspectives on activity theory. Y. Engeström, R. Miettinen and R. L. Punamäki (eds), 377–404. Cambridge: Cambridge University Press. Engeström, Yrjö 2004 ‘New forms of learning in coconfiguration work’. Journal of Workplace Learning 16/1&2: 11–21. Engeström, Yrjö 2006a ‘Activity theory and expansive design’ in Theories and practice in interaction design. S. Bagnara and G. Crampton Smith (eds), 3–24. Mahwah, NJ: Lawrence Erlbaum. Engeström, Yrjö 2006b ‘From well-bounded ethnographies to intervening in mycorrhizae activities’. Organization Studies 27/12: 1783–1793.
Engeström, Yrjö, and Frank Blackler 2005 ‘On the life of the object’. Organization 12/3: 307–330. Engeström, Yrjö, Katherine L. Brown, Carol Christopher, and Judith Gregory 1997 ‘Coordination, cooperation, and communication in the courts: Expansive transitions in legal work’ in Mind, culture, and activity: Seminal papers from the laboratory of comparative human cognition. M. Cole, Y. Engeström and O. Vasquez (eds), 369–385. Cambridge: Cambridge University Press. Faraj, Samer A., and Lee S. Sproull 2000 ‘Coordinating expertise in software development teams’. Management Science 46/12: 1554–1568. Feller, Joseph, and Brian Fitzgerald 2001 Understanding open source software development. London: Addison-Wesley. Goulding, Christina 2002 Grounded theory: A practical guide for management, business and market researchers. London: Sage. Hemetsberger, Andrea, and Christian Reinhardt 2006 ‘Learning and knowledge-building in open-source communities: A social-experiential perspective’. Management Learning 37/2: 187–214. Hertel, Guido, Sven Niedner, and Stefanie Herrmann 2003 ‘Motivation of software developers in open source projects: An internetbased survey of contributors to the Linux kernel’. Research Policy 32/7: 1159–1177. Hutchins, Edwin 1993 ‘Learning to navigate’ in Understanding practice: Perspectives on activity and context. S. Chaiklin and J. Lave (eds), 35–63. Cambridge: Cambridge University Press. Hutchins, Edwin 1995 Cognition in the wild. Cambridge: MIT Press. Kaptelinin, Victor 1996 ‘Computer-mediated activity: Functional organs in social and developmental contexts’ in Context and consciousness: Activity theory and human computer interaction. B.A. Nardi (ed.), 23–34. Cambridge: MIT Press.
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
Hemetsberger & Reinhardt Collective Development in Open-Source Communities Kellogg, Katherine C., Wanda J. Orlikowski, and JoAnne Yates 2006 ‘Life in the trading zone: Structuring coordination across boundaries in postbureaucratic organizations’. Organization Science 17/1: 22–44. Kogut, Bruce, and Anca Metiu 2001 ‘Open-source software development and distributed innovation’. Oxford Review of Economic Policy 17/2: 248–264. Kozinets, Robert V. 2002 ‘The field behind the screen: Using netnography for marketing research in online communities’. Journal of Marketing Research 39/2: 61–72. Kreiner, Kristian, and Majken Schultz 1993 ‘Informal collaboration in R & D: The formation of networks across organizations’. Organization Studies 14/2:189–209. Kuk, George 2006 ‘Strategic interaction and knowledge sharing in the KDE developer mailing list’. Management Science 52/7: 1031–1042. Kuutti, Kari 1996 ‘Activity theory as a potential framework for human–computer interaction research’ in Context and consciousness: Activity theory and human computer interaction. B.A. Nardi (ed.), 17–44. Cambridge: MIT Press. Lanzara, Giovan F., and Michele Morner 2003 ‘The knowledge ecology of opensource software projects’. Paper presented at the 19th EGOS Colloquium (July), Copenhagen. Lanzara, Giovan F., and Michele Morner 2005 ‘Artefacts rule! How organizing happens in open source software projects’ in Actor network theory and organizing. B. Czarniawska and T. Hernes (eds), 67–90. Copenhagen: Copenhagen Business School Press. Latour, Bruno 1996 ‘Social theory and the study of computerized work sites’ in Information technology and changes in organizational work. W. J. Orlikowski, G. Walsham, M. R. Jones and J. I. DeGross (eds), 295–307. London: Chapman & Hall. Lave, Jean, and Etienne C. Wenger 1991 Situated learning: Legitimate peripheral participation.
1007
Cambridge: Cambridge University Press. Lee, Gwendolyn K., and Robert E. Cole 2003 ‘From a firm-based to a communitybased model of knowledge creation: The case of the Linux kernel development. Organization Science 14/6: 633–64. Leontiev, Alexei N. 1978 Activity, consciousness, and personality. Englewood Cliffs, NJ: Prentice-Hall. Leontiev, Alexei N. 1981 ‘The problem of activity in psychology’ in The concept of activity in Soviet psychology. J. Wertsch (ed.). Armonk, NY: Sharpe. Metiu, Anca 2006 ‘Owning the code: Status disclosure in distributed groups’. Organization Science 17/4: 418–435. Nardi, Bonnie A., (ed.) 1996 Context and consciousness: Activity theory and human computer interaction. Cambridge: MIT Press. Nardi, Bonnie A., Steve Whittaker, and Heinrich Schwarz 2000 ‘It’s not what you know, it’s who you know: Work in the information age’. First Monday 5/5 http://firstmonday. org/issues/issue5_5/nardi/index.html Orlikowski, Wanda J. 2000 ‘Using technology and constituting structures: A practice lens for studying technology in organizations’. Organization Science 11/4: 404–428. Orlikowski, Wanda J. 2002 ‘Knowing in practice: Enacting a collective capability in distributed organizing’. Organization Science 13/3: 249–273. Raymond, Eric S. 1999 The cathedral and the bazaar: Musings on Linux and open source by an accidental revolutionary. Sebastopol, CA: O’Reilly and Associates. Sapsed, Jonathan, and Ammon Salter 2004 ‘Postcards from the edge: Local communities, global programs and boundary objects’. Organization Studies 25/9: 1515–1534. Suchman, Lucy 1987 Plans and situated actions: The problem of human–machine communication. Cambridge: Cambridge University Press.
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010
1008
Organization Studies 30(09) Tuomi, Illka 2001 ‘Internet, innovation, and open source: Actors in the network’. First Monday 6/1 http://www. firstmonday. org/issues/issue6_1/ tuomi/index.html Turkle, Sherry 1995 Life on the screen: Identity in the age of the internet. New York: Simon & Schuster. van der Hoek, André, David Redmiles, Paul Dourish, Anita Roberto, Sarma Silva Filho, and Cleidson de Souza 2004 ‘Continuous coordination: A new paradigm for collaborative software engineering tools’ in Proceedings of the Workshop on WoDISEE Directions in Software Engineering Environments, 29–36. Victor, Bart, and Andrew C. Boynton 1998 Invented here: Maximizing your organization’s internal growth and profitability – A practical guide to transforming work. Boston: Harvard Business School Press. von Hippel, Eric, and Georg von Krogh 2003 ‘Open source software and the ‘private-collective’ innovation model: Issues for organization science’. Organization Science 14/2: 209–223. von Krogh, Georg, and Eric von Hippel 2006 ‘The promise of research on open source software’. Management Science 52/7: 975–983.
von Krogh, Georg, Sebastian Spaeth, and Karim R. Lakhani 2003 ‘Community, joining, and specialization in open source software innovation: A case study’. Research Policy 32/7: 1217–1241. Vygotsky, Lev S. 1978 Mind in society. Cambridge, MA: Harvard University Press. Vygotsky, Lev S. 1981 ‘The instrumental method in psychology’ in The concept of activity in Soviet psychology. J. Wertsch (ed.), 134–143. Armonk, NY: Sharpe. Wells, Gordon 2002 ‘The role of dialogue in activity theory’. Mind, Culture, and Activity 9/1: 43–66. Yamauchi, Yutaka, Makoto Yokozawa, Takeshi Shinohara, and Toru Ishida 2000 ‘Collaboration with lean media: How open-source software succeeds’. Proceedings of the 2000 ACM conference on Computer Supported Cooperative Work, 329–338. Zeitlyn, David 2003 ‘Gift economies in the development of open source software: Anthropological reflections’. Research Policy 32/7: 1287–1291.
Andrea Hemetsberger
Andrea Hemetsberger is Associate Professor at University of Innsbruck School of Management, Austria. She holds a PhD in Marketing from the University of Innsbruck, and has visited Tilburg University in The Netherlands, the Schulich School of Business, Canada, and ESSEC Business School, and Université Paris-Dauphine, France. Her research interests revolve around free open-source software research, branding, consumer devotion and creative consumers. She has published in the Journal of Business Research, Management Learning, Journal of Business-to-Business Marketing, Advances in Consumer Research and the Journal of Macromarketing. Address: Department of Strategic Management, Marketing and Tourism, University of Innsbruck School of Management, Universitaetsstrasse 15, 6020 Innsbruck, Austria. Email: andrea.hemetsberger@uibk.ac.at
Christian Reinhardt
Christian Reinhardt is a PhD student at University of Innsbruck School of Management, Austria, and independent consultant in the area of (online) community-building and brand management. His research focuses on online collaboration and knowledge creation in free open-source software communities. He has published in Management Learning, and has presented his research in books and at various conferences. Address: University of Innsbruck School of Management, Universitaetsstrasse 15, 6020 Innsbruck, Austria. Email: christian.reinhardt@uibk.ac.at
Downloaded from oss.sagepub.com by Giorgio Bertini on October 16, 2010