Int. J. Technology Management, Vol. 53, No. 1, 2011 69

Copyright © 2011 Inderscience Enterprises Ltd.

Thinking along: a process for tapping into knowledge across boundaries

Hans Berends* School of Industrial Engineering, Eindhoven University of Technology, Pav.L.03, P.O. Box 513, 5600 MB Eindhoven, The Netherlands Fax: +31-40-246-8054 E-mail: j.j.berends@tue.nl *Corresponding author

Raghu Garud SMEAL College of Business, Pennsylvania State University, 431 Business Building, University Park, PA 16802, USA E-mail: rug14@smeal.psu.edu

Koenraad Debackere Katholieke Universiteit Leuven, Naamsestraat 69, Leuven 3000, Belgium E-mail: koenraad.debackere@econ.kuleuven.be

Mathieu Weggeman School of Industrial Engineering, Eindhoven University of Technology, Pav.M.09, P.O. Box 513, 5600 MB Eindhoven, The Netherlands E-mail: m.c.d.p.weggeman@tue.nl

Abstract: The knowledge management literature offers knowledge ‘transfer’ and ‘transformation’ as mechanisms to tap into the diversity of knowledge that lies dispersed over organisations. However, it is difficult to transfer and transform knowledge across epistemic boundaries and, even if these processes were to unfold, specialisation could well be compromised. Based on in-depth studies carried out within two industrial research organisations, we describe an alternative mechanism for tapping into knowledge across boundaries. We call this ‘thinking along’, an interactive process that allows a person with a problem to tap into someone else’s knowledge base without them having to get involved in each others’ ways of knowing. In contrast to knowledge transfer and knowledge transformation, thinking along circumvents interpretative barriers, while retaining specialisation.

70 H. Berends et al.

Keywords: industrial research; knowledge management; knowledge sharing; communities-of-practice; organisational knowledge; specialisation.

Reference to this paper should be made as follows: Berends, H., Garud, R., Debackere, K. and Weggeman, M. (2011) ‘Thinking along: a process for tapping into knowledge across boundaries’, Int. J. Technology Management, Vol. 53, No. 1, pp.69–88.

Biographical notes: Hans Berends is an Assistant Professor in the Innovation, Technology Entrepreneurship and Marketing Group of the School of Industrial Engineering at Eindhoven University of Technology (The Netherlands). He holds a PhD from Eindhoven University of Technology for a thesis on knowledge sharing in industrial research. He has published on knowledge sharing and organisational learning in Human Relations, Journal of Product Innovation Management, Social Epistemology and R&D Management. His current research interests focus on process dynamics of innovation and learning.

Raghu Garud is Alvin H. Clemens Professor of Management and Organisation and the Research Director of the Farrell Center for Corporate Innovation and Entrepreneurship, Pennsylvania State University. His research currently explores the emergence of novelty. Specifically, he is interested in understanding how new ideas emerge, are valued and become commercialised, offering concepts such as path creation, technology entrepreneurship and bricolage as a collective process. His most recent article in Organization Science explores the role that narratives play in sustaining innovation within organisation. He was, till recently, an Associate Editor of Management Science and Co-editor of Organization Studies.

Koenraad Debackere is a Professor of Managerial Economics, Strategy and Innovation at Katholieke Universiteit Leuven (Belgium). He obtained his PhD in Management with an ICM-Fellowship at the University of Gent after his stay as an ICM-Fellow and an ICRMOT Research Assistant at MIT Sloan School of Management. He was a Fulbright-Hays Postdoctoral Fellow at MIT in 1991 to 1992. His research has focused on the area of technology and innovation management, the development of indicators for measuring the linkage between science and technology, the design and use of bibliometric indicators for science policy purposes and the role of entrepreneurial universities in economic development.

Mathieu Weggeman is Professor of Organisation Science and Innovation Management in the School of Industrial Engineering at Eindhoven University of Technology (The Netherlands). He received his PhD in Strategic Management from Tilburg University. His primary expertise is in the field of organisational knowledge creation in the early stages of the innovation process. A second area of his interest concerns the design of organisations in which professionals are motivated to high performance. He has published books and articles in the fields of participative strategy development, knowledge management in professional organisations and the concept of knowledge.

1 Introduction

The importance of knowledge management in R&D is well-established (Frederiksen et al., 2004; Rui et al., 2008). Yet, R&D organisations face a dilemma with regard

Thinking along 71

to knowledge management. On one hand, organisations benefit from specialisation (Grant, 1996b; Postrel, 2002; Sanchez and Mahoney, 1996). The enormous growth and sophistication of technological knowledge requires that individuals specialise to enable organisations to acquire, develop and use state-of-the-art knowledge in a wide range of fields (Boland and Tenkasi, 1995; Carlile, 2002). On the other hand, organisations need to tap into the diversity of knowledge that lies distributed across organisation members so as to facilitate knowledge reuse (Argote, 1999; Majchrzak et al., 2004) and enable innovation through the recombination of ideas (Leonard-Barton, 1995; Galunic and Rodan, 1998; Garud and Nayyar, 1994).

Prior research has proposed knowledge transfer as a mechanism to leverage knowledge that lies dispersed within organisations (Argote and Ingram, 2000; Levin and Cross, 2004). Yet, epistemic boundaries that arise between knowledge domains hamper knowledge transfer (Brown and Duguid, 2001; Knorr-Cetina, 1999; Dougherty, 1992). Other studies suggest knowledge transformation as a mechanism to deal with these epistemic boundaries between specialists (Bechky, 2003; Carlile, 2004). However, transformation is difficult as it requires the development of in-depth mutual understanding and the reconciliation of different perspectives. Moreover, both knowledge transfer and transformation make knowledge directly available to other persons to create common knowledge, thereby, reducing the benefits from specialisation (Demsetz, 1988; Grant, 1996a, 1996b).

How might organisation members tap into the diversity of knowledge that lies distributed across an organisation without having to transfer or transform knowledge? We address this question through field studies in two industrial research organisations. We describe a mechanism that we found that these organisations used that we label as ‘thinking along’. Thinking along is an interactive process between two or more parties – one that has a problem and another who contributes towards its solution – wherein a solution to the problem emerges without the participants having to become involved in each others’ ways of knowing.

Based on an in-depth analysis of an extensive set of thinking along episodes, we flesh out several interrelated elements of the process: initiating, connecting, contributing and concluding. Our analysis of thinking along processes shows its distinctive features vis-à-vis knowledge transfer and knowledge transformation. We reflect on these issues in the discussion section of the paper and provide a summary of our theoretical contributions in the conclusion section of this paper. Specifically, we revisit how thinking along makes it possible for individuals to benefit from the knowledge possessed by others while retaining specialisation.

2 Theoretical background

Over the past decade, a practice-based perspective on knowing and learning in organisations has emerged (Amin and Roberts, 2008; Gherardi, 2006). The central thesis that has emerged from a range of ethnographic studies is that knowing is embedded in practice (Carlile, 2002; Lave and Wenger, 1991; Nicolini et al., 2003). Such a perspective is appreciative of the fact that performance unfolds within a material and social context and that such a context gives meaning and structure to what is being done [Wenger, (1998), p.47]. The technicians studied by Orr (1990), for example, practiced their

72 H. Berends et al.

repair work in the context of photocopying machines, spare parts, customers, colleagues, documentation and stories. Knowledge is embedded in these practices and their material and social environments (Gherardi, 2006; Lave and Wenger, 1991). For instance, tools and resources are essential as they provide actors with the necessary information and clues to act knowledgeably (Tyre and Von Hippel, 1997). Similarly, knowledge is embedded in a social environment because social interactions determine what counts as knowledge and because it is in collaboration with others that people are able to act knowledgeably (Latour, 1987; Hutchins, 1995).

Groups that are centred on a set of practices develop shared ways of working, knowing and learning (Boland and Tenkasi, 1995). Tightly knit groups who participate in the same practice form a community-of-practice (Lave and Wenger, 1991; Brown and Duguid, 1991). Although not all knowledge will be shared among those who participate in a common practice, the shared background of community members enables easy knowledge transfer (Brown and Duguid, 2001; Wenger, 1998). As Orr (1990, p.170) pointed out, ‘the common experience of practice provides the context which makes meaningful those points which could never be explained without the experience’.

R&D staff members in larger organisations usually engage in a heterogeneous set of practices and have to develop competences in associated knowledge domains. For example, an automobile company not only requires competence in vehicle and engine production, but also in many other major technological fields including chemical processing, metallurgy, semiconductors, instruments and controls (Granstrand et al., 1997). To build such a differentiated knowledge base, staff members focus on few practices and specialise around different problems (Boland and Tenkasi, 1995; Carlile, 2002). This specialisation enables individual researchers to master state-of-the-art expertise, thereby, enabling the organisation to obtain both the range and the quality of knowledge required for complex innovation processes (Carlile, 2002; Granstrand et al., 1997). Consequently, an organisation becomes a distributed knowledge system in which knowledge is dispersed across its members (Boland and Tenkasi, 1995; Tsoukas, 1996).

Innovative activities, though, benefit from organisation members drawing upon the diversity of knowledge that lies dispersed over the organisation (Allen, 1977; Garud and Nayyar, 1994; Galunic and Rodan, 1998; Miller et al., 2007). One strategy proposed to tap into distributed knowledge is knowledge transfer. Knowledge transfer is defined as the transmission and receipt of knowledge (Grant, 1996b; Szulanski, 2000) – a process that is most easily accomplished across actors who share a common practice (Boland and Tenkasi, 1995). A consequence of the specialisation of groups of organisation members is that epistemic boundaries emerge between domains of practice (Carlile, 2002; Wenger, 1998). These epistemic boundaries consist of differences in the ways of knowing associated with domains of practice. Practices may differ in conceptual schemas, procedures for measurement and experimentation, the use of artefacts and the social organisation of knowledge production (Knorr-Cetina, 1999). These differences hamper the creation of mutual understanding across boundaries as people from different communities lack common points of reference to make sense of each other’s practices. Knowledge transfer is thus hampered by a lack of shared meanings (Bechky, 2003).

Knowledge transformation is another process that organisations use to tap into distributed knowledge. Majchrzak et al. (2004) found that knowledge reuse for radical innovation required the transformation of concepts taken from other contexts, not the literal replication of knowledge as assumed in the literature on knowledge transfer. Bechky (2003) observed how knowledge from one community was made to fit in the

Thinking along 73

work context of another community through a process of knowledge transformation. In this vein, Carlile and Rebentisch (2003) described knowledge transformation as a process of identifying and representing the knowledge of different groups involved in a project and renegotiating the value of each other’s perspective. Clearly, the creation of mutual understanding and the reconciliation of differences may be a difficult process (Bechky, 2003; Carlile, 2004).

Besides the difficulties of transferring and transforming knowledge across epistemic boundaries, both processes are at odds with specialisation (Demsetz, 1988; Grant, 1996b). An organisation that requires each individual to learn what every other individual knows is inherently inefficient (Grant, 2001). Unrestrained knowledge transfer and transformation can create cognitive overload for individuals (Grant, 1996b). Such overload reduces the variety of knowledge that an organisation may possess. In turn, this reduces its adaptive capacity because it will not possess the requisite variety of knowledge (Ashby, 1956) to cope with the complexities it confronts (Garud and Kumaraswamy, 2005).

3 Thinking along

Is there a way to tap into knowledge that lies distributed across an organisation without the need for transfer or transformation? Our study of episodes in two research sites – Philips’ NatLab and the Oil and Gas Innovation Research (OGIR) Group of Shell Global Solutions – suggests an approach that we label as ‘thinking along’. Consider an episode between two researchers – Richard and Andrew – both members of the Buijs Group at the NatLab. Richard was a young research engineer, a physicist and member of the Coatings Cluster of the Buijs Group. He experimented with spin coating a polymer layer on optical discs as part of a project funded by a product division. Spin coating is a process whereby a thin and homogeneous layer is created by dropping liquid on a disc and rotating it at a very high frequency. This practice occurs within a material environment including spinning equipment, coating liquids, curing techniques, dust management, and instruments to measure layer thickness and flatness. Experiments typically vary parameters like spinning frequency, spinning-time and ramp up time in order to arrive at the required thickness and homogeneity without defects.

At one point, Richard confronted a problem – the results from spin coating differed from disc to disc: ‘I was getting a strange phenomenon. Using the same parameters every time … the layer was slowly getting thinner and thinner. I couldn’t work it out’. Richard wondered whether the problem might have been caused by changes in the temperature of the polymer liquid he had used, but because he was a physicist, he knew little about polymers. He sought out Andrew, a chemical engineer from another cluster in the Buijs Group. Richard had not interacted with Andrew before, but he knew that Andrew was working on polymer processing. Richard explained his problem and proposed his hypothesis about the influence of temperature on the viscosity of the polymer liquid. Andrew rejected this hypothesis because he did not believe that the viscosity of the polymer could be affected that strongly by temperature. ‘But there’s something else’, he said. Andrew hypothesised that it could very well be that the polymer liquid, which was stored in a glass bottle, opened up and formed radicals under the influence of UV light. If that were the case, the radicals would react with oxygen every time Richard opened the

74 H. Berends et al.

bottle. The longer the bottle remained open, the more the liquid changed and the more this influenced the spin coating process. ‘That was a breakthrough. I hadn’t even thought about it’, said Richard. Richard tested the explanation by changing the way he treated the bottles. This adjustment proved to be effective as it solved his problem.

Richard’s theories, rules of thumb and practical skills associated with his equipment and the spin coating parameters proved to be insufficient for dealing with the problem. Therefore, he approached someone who was engaged in another practice that seemed relevant for solving his problem. Andrew was a chemical engineer and held a PhD on polymer processing. At the time of our study, Andrew was involved in a project on the bonding of polymers and metals. Both his educational background and his day-to-day practice differed strongly from Richard’s. Andrew analysed problems in terms of molecular chains, activation energies and chemical reactions. Yet, the value of the episode was precisely that Andrew approached the situated problem from his own perspective. Andrew did not know about Richard’s problem in advance, let alone have an explanation for its occurrence. Andrew applied his knowledge of polymer chemistry to develop an ‘outsider’s’ understanding of Richard’s problem, to assess Richard’s hypothesis and to put forth an alternative explanation. The novel hypothesis was communicated to Richard. Andrew did not need to understand Richard’s entire practice, nor did he not have to get involved in spin coating.

Conversely, Richard did not learn much about polymers because Andrew did not attempt to communicate all relevant polymer knowledge. Richard did not understand the complete details as to how and why Andrew had arrived at his conclusion, nor was there any need to. Indeed, it would be hard for Richard to have incorporated into his day-to-day practice Andrew’s background knowledge. Such knowledge did not fit in with the material context of Richard’s work, nor with his existing knowledge, or with the social context of his practices (since the project plan did not leave room for a crash course on polymers).

We adopted the expression used by some group members who labelled such interactions as ‘thinking along’ with someone. We define thinking along as an interactive process that allows a person with a problem to tap into another person’s knowledge base without having to get involved in each others’ ways of knowing. Thinking along was a widespread practice in both research organisations that we studied, occurring in about one-third of the episodes that we identified. Therefore, we concluded that it was a significant phenomenon warranting further investigation.

4 Research sites and methods

Our empirical research consisted of qualitative field studies in two industrial research groups. The R&D context is particularly appropriate for this study as R&D requires both a high degree of specialisation (Granstrand et al., 1997; Pavitt, 1998) and, at the same time, considerable collaboration across boundaries (Garud and Nayyar, 1994). This is because research challenges are unpredictable and, so, researchers frequently tap into other’s knowledge (Ravasi and Verona, 2001; Van Looy and Debackere, 2001). We focus in particular on interactions among researchers who are engaged in different practices and are not involved in joint projects because the crossing of boundaries makes knowledge transfer difficult (Brown and Duguid, 2001) and the lack of task

Thinking along 75

interdependence reduces the need to develop shared understanding through knowledge transformation (Carlile, 2004).

4.1 Research sites

As we mentioned earlier, the ‘NatLab’, the main research organisation of Royal Philips Electronics1, was one of our research sites. The NatLab is one of the largest industrial research laboratories in the world, as it employs around 1,700 researchers, working in many specialised domains. From its inception in 1914, the NatLab has been located in Eindhoven (The Netherlands). In particular, we focused on members of the Buijs Group, which consisted of about 25 researchers at the time of study. Two-thirds were research scientists holding PhDs in physics, chemistry and mechanical engineering or related fields. The other third were research engineers with a higher technical or laboratory-oriented education. The Buijs Group’s activities were divided into six clusters: ‘solid mechanics and tribology’, ‘plastic processing’, ‘thermal management’, ‘coating’, ‘printing’ and ‘home care’.

To increase the scope and variety of observations, we conducted a second field study at OGIR. OGIR is a business group within Shell Global Solutions International and is located in Amsterdam, The Netherlands. Shell Global Solutions is part of the Royal Dutch/Shell Group of companies. OGIR described its mission as contributing to sustainable development in the areas of energy and mobility by generating innovative technological options. At the time of our study, this group consisted of about 30 members and its composition was comparable to that of the Buijs Group.

Both research groups have been engaged in fundamental research with researchers regularly publishing articles in scientific and technical journals. The primary objective of both groups is to deliver new technological options that can be developed further by other groups. Both the Buijs Group and OGIR organise work in projects, which are funded by business divisions or by a company budget for fundamental research. On average, the projects within both groups are staffed with less than two researchers. A project team often consists of a researcher and a research assistant or a research engineer. These projects depend strongly on the knowledge of colleagues close at hand and faraway.

At the field study sites, the boundaries between practices cut across groups, projects and departments. At the NatLab, for example, researchers belong to a group (e.g., the Buijs Group), to a specific cluster within that group (e.g., coating) and to a project team (e.g., PolyLed spin coating). Moreover, they are characterised by different scientific backgrounds (e.g., chemical engineering or physics) and by position (scientific staff, research engineer or trainee). Within each organisational sub-unit, people are likely to share some but not all practices. Members of the coating cluster, for example, share the practice of studying the fluid dynamics of coating processes. This means that they share theoretical concepts (e.g., ‘viscosity’, ‘Marangoni number’), are jointly experimenting with different coating processes (e.g., ‘spin coating’, ‘flow coating’, ‘spray coating’) and have a common way of observing defects in coatings (e.g., ‘oak leaves’, ‘beach effect’). Yet, members of one cluster do not share all practices. One researcher may be engaged in numerical simulation and might share this practice with non-cluster members, but not with members of the same cluster who are engaged primarily in experimental work or analytical research.

76 H. Berends et al.

4.2 Data collection

The field studies can be classified as passive participant observation (Spradley, 1980). One member of our research team temporarily shared a room with different researchers, followed them to meetings and to their laboratories, and joined them for coffee and lunch breaks. However, he did not actively participate in their research. During the first phase of the field studies, interviews were held with most of the group members. These interviews served not only as a source of factual information, but also as an opportunity to establish mutual trust and to negotiate further access.

To facilitate access to interactions and to gain insight into the researchers’ contexts, we chose to focus on the interactions of a limited number of researchers. We shadowed six researchers for several days (see Table 1). The interactions we observed comprised group meetings, cluster meetings, project meetings, research colloquia, appointments between individual researchers, lunches and coffee breaks, and informal meetings in the corridor. The episodes were tape recorded where possible. Where tape recording was deemed to be disruptive (such as a spontaneous encounter in the corridor), we took notes. We considered each of these meetings and interactions as an episode and gave each episode a number label (e.g., E26). Before and after interactions, the individuals who had been shadowed were asked to explain the meaning of their interactions. In many cases, we also interviewed the others involved in the interactions to learn their point of view. A number of the tape recorded interactions were discussed on a sentence-by-sentence basis with the respondents. In addition to the face-to-face interactions, we analysed a few written exchanges and telephone conversations. At the end of the field studies, we presented preliminary findings to both research groups (Lincoln and Guba, 1985). These member checks did not require any major revisions. Table 1 The characteristics of the field studies

Buijs Group OGIR

Period April–October 1999 March–September 2001 Days spent at research group 45 31 Introductory interviews conducted 22 23 Researchers ‘shadowed’ 4 2 Days spent ‘shadowing’ researchers 19 8

4.3 Data analysis

We followed qualitative procedures to analyse the field notes and transcripts (Strauss and Corbin, 1990; Miles and Huberman, 1994). Through systematic comparison, we started coding the issue addressed in the episodes (e.g., whether or not it was someone’s specific problem), the roles of the actors involved (e.g., problem owner or contributor) and whether epistemic boundaries were crossed (i.e., when a contributor approached a problem from a domain of practice that was not familiar to the problem owner). This latter issue was assessed based upon participants’ comments during and after the conversations referring to a lack of expertise in a domain and our own knowledge of participants’ activities and group membership.

Thinking along 77

Then, we focused on the knowledge dynamics involved. While several episodes resembled processes of knowledge transfer or transformation, others showed a different pattern. Using a concept used by the researchers themselves, we called these interactions ‘thinking along’. To get a sense of the importance of the phenomenon, we analysed 109 interaction episodes that we had identified (52 from OGIR and 57 from the Buijs Group, taken from five successive weeks). To be coded as ‘thinking along’, an episode had to be interactive and focused on a specific problem of one of the actors, to which another person contributed without both parties becoming deeply involved in each other’s knowledge bases. Out of the 109 episodes, 36 could completely or partly be coded as thinking along. In most of these cases, clear epistemic boundaries were crossed, but in some cases, this was a matter of degree as the lines of demarcation between practices were not sharp (cf. Wenger, 1998).

To further investigate the process of thinking along, we followed Boland and Tenkasi’s (1995) suggestion to use a pragmatic model of communication to study knowledge processes across boundaries instead of the conduit model of senders and receivers. Informed by speech act theory (Searle, 1969), we use the notion of a move as a meaningful unit of communication [Goffman, (1981), p.24; Pentland, 1992]. A move is a turn or a part of a turn in which a particular act is performed, like making a promise, defending an argument or opening a meeting. The analysis of moves helped to flesh out the subprocesses of thinking along.

Figure 1 Thinking along subprocesses

Initiating Problem descriptions

Questions

Concluding Evaluations

Suggestions

Connecting Contributing

Contributor

Problem owner

78 H. Berends et al.

Table 2 Examples of thinking along episodes

Initiating Connecting Contributing Concluding

E37: Jason and Marc

A chemical engineer from another group at the NatLab, Marc, approaches the theoretical physicist Jason, member of the Buijs Group. They know each other from joint involvement in prior projects.

Marc introduces his problem that a particular pattern of irregularities is formed in a coating process. Jason, in turn, moves back and forth from questions to potential explanations.

Jason forms a hypothesis on the cause of the problem: water might exist in the coating liquid. He also thinks up a solution. Marc himself also comes up with a supplementary explanation for his observations.

The conversation shifts to a problem faced by Jason on which Marc provides unsollicited advice, which is dismissed by Jason. In the end, Jason provides Marc with a bottle of liquid to overcome the irregularities.

E44: Richard and Andrew

Richard initiates an interaction with Andrew, a chemical engineer from another cluster in the Buijs Group. He had not interacted with Andrew before, but knew that Andrew was working on polymer processing.

Richard describes unstable results from a spin coating process. He proposes a hypothesis about the influence of temperature on the viscosity of the polymer liquid. Andrew asks additional questions.

Andrew rejects Richard’s original hypothesis and offers another hypothesis focusing on the polymer liquid forming radicals under the influence of UV light. Together, they explore potential solutions.

Richard leaves the episode with Andrew’s, yet untested, hypothesis. Later, he tested the explanation by changing the way he treated the bottles (which proved to solve the problem).

E69: Jason and Luke

Luke, a Buijs Group researcher, approaches Jason for his expertise in optics and optical filters, a domain in which Luke is not active. Luke has interacted with Jason about the issue before and builds upon earlier conversations.

Luke explains that he got distorted pictures from an infrared camera and wonders whether the noise is caused by the optical filter used. He shows a graph of the filter’s characteristics. Jason asks for additional information and why this is so important.

Jason is reluctant to provide an answer. After additional questions by Luke, Jason applies his knowledge of optics and deduces from the shape of the curve that the filter’s performance seems to be sufficient, but adds that he cannot guarantee certainty.

For Luke, this conclusion suffices. He does not understand the complete background of Jason’s conclusion, but knowing Jason’s high scientific standards, he says: ‘now, I am pretty sure that I am not fooled by the supplier of the filter’.

Thinking along 79

Table 2 Examples of thinking along episodes (continued)

Initiating Connecting Contributing Concluding

E209: Malcolm and OGIR researchers

Malcolm presents a potential research programme concerning CO2 sequestration to 15 senior OGIR researchers at a lunch meeting, all involved in different specialties within chemical engineering.

Malcolm describes a well-known but very slow chemical reaction and asks his colleagues for ideas to speed up the process and to overcome other difficulties. After the researchers have provided their ideas, he presents his own ideas and asks for comments.

Although these questions are new to his colleagues, they indeed come up with two possible solutions. Later, he also gets feedback on his own proposed solutions.

Two solutions raise the enthusiasm of Malcolm and are further explored to turn into potential projects.

Table 3 Conditions for thinking along

Initiating Connecting

• Meta-knowledge about others

• Creation of opportunities

• Problem owner willing to acknowledge problems

• Contributor willing to take time and use knowledge

• Trust and commitment

• Active engagement of both problem owner and contributor

• Problem owner frames problem comprehensibly, possibly supported by boundary objects

• Potential contributor is able to make sense of problem from his or her perspective

• Basic level of shared understanding

Contributing Concluding

• Both problem owner and contributor actively involved in suggestions and evaluations

• Problem owner is open for contributions from other perspectives

• Return to activity of connecting when noticing a disconnect in how the problem owner and the contributor view the problem

• Not necessarily striving for consensus or forcing closure

• Using conversation to increase meta-knowledge and enable future interactions

5 The process of thinking along

By content analysing the data, we emerged with four interrelated activities that comprise thinking along: initiating, connecting, contributing and concluding (Figure 1). Table 2 presents these activities in four examples of thinking along episodes. Although these

80 H. Berends et al.

activities form a logical sequence, these activities are iterative, rather than linear. The core of the process can be conceived of as iterative cycles of connecting and contributing, in which the problem owner and a contributor are interactively engaged. Further, episodes of thinking along often build upon earlier episodes and their conclusion does not necessarily mark the end of the engagement with each others’ work. Our discussion of these different activities also leads to the identification of conditions for thinking along (Table 3).

5.1 Initiating

Although written media sometimes played a role in thinking along, most of it occurred during face-to-face encounters. A key condition for thinking along was the existence or creation of such an opportunity for interaction. At the NatLab, group work meetings institutionalised opportunities for thinking along. At these biweekly meetings, one group member presented a piece of research in progress. This offered other group members, mostly working in different specialties, with the opportunity to respond to preliminary conclusions and open questions. Frequently, exchanges that started within an organised meeting were continued informally. At OGIR, this practice was less institutionalised, although some researchers took the initiative for monthly lunch meetings with a similar aim (e.g., E209 in Table 2).

Most cases of thinking along, though, occurred within informal meetings, usually initiated by the problem owner, like the interaction between Richard and Andrew. Meta-knowledge on who knows what enabled problem owners to turn to the colleagues who master a particular practice (cf. Akgün et al., 2006; Brandon and Hollingshead, 2004). Both at OGIR and at the NatLab group meetings and colloquia were organised with the explicit intention to increase this knowledge about others. Furthermore, interactions provide employees with an opportunity to learn about the capabilities of others and the types of problems they are able to solve. Chris, a research engineer at the Buijs Group, said: ‘when I seem to be just chatting somewhere, that chatting is purposeful: to stay informed about what my colleagues know!’ Such interactions also build trust and commitment. One of the researchers of the Group Buijs said: ‘a lot of work is done at the corridors. (…) Partly that is gossiping. ‘I tell you something and therefore you are my friend’. The function of that is to build trust. That makes it possible to go and ask for advice later’. Episodes among previously unacquainted researchers usually started by each of them describing the organisational and project-related background of their work. Other episodes could build upon previous conversations and sometimes started ‘in media res’.

Finally, initiation depended on problem owners to acknowledge a need for input from others and on the willingness of contributors to spend time and share ideas. The former condition proved to be a problem more often than the latter. While literature on knowledge management literature often suggests that professionals are reluctant to ‘give away’ their knowledge (cf. Andrews and Delahaye, 2000; Shih et al., 2006), such a threat did not play a major role in thinking along processes. In thinking along, much knowledge is used, but only a fraction of it is given away. Researchers actually liked being creatively involved in each other’s work and no mention was made of any unwillingness to think along for a moment. Jason commented after a meeting with Luke (E69; see Table 2): ‘if someone comes to my office, I will lay aside the work that I am busy with at that moment. If someone calls by phone I will make an appointment. But usually, they drop by, like

Thinking along 81

‘could you think along with me for a minute’. They often come with interesting questions. Like Luke just did’.

On the contrary, researchers were sometimes reluctant to ask for assistance because asking others could be perceived as a weakness (cf. Lee, 1997). Letting others think along implies the acknowledgement of a problem to others. Luke explained his decision to approach Jason with a question about an optical filter: ‘it is possible to draw conclusions from such a graph, but I lack the expertise to do so. But someone familiar with these optical filters should be able to judge its quality based on this graph’ (E69; see Table 2). For some researchers, such an acknowledgement posed a barrier. Andrew said: ‘I will always try to solve the problem by myself first. If that fails, I have to choose between keeping my pride and having a quick answer’.

5.2 Connecting

The second key element of thinking along is connecting. Problem owners need to connect with another knowledge domain and potential contributors need to connect with the problem by making sense of it from their perspective. This bidirectional activity of connecting is thus a limited form of perspective taking (Boland and Tenkasi, 1995). Connecting includes two types of moves that help potential contributors to make sense of other persons’ problems: problem descriptions and questions. In most episodes, problem owners started by introducing a problem that they were struggling with. Usually, it was not considered sufficient just to ask a question. Problem owners often did not have very specific ideas about potential answers and potential contributors needed more cues on the context of the problem to approach it effectively from their own perspective. Problem owners described the context of the problem, symptoms (‘the layer is getting thinner and thinner’), objectives (‘we need to speed up this process’) and solutions that had been tried before. Problem owners tried to phrase these issues in a way that made sense to the potential contributor, thereby, resembling other kinds of help seeking behaviour (cf. Constant et al., 1996; Lee, 1997). This was often supported by using boundary objects (cf. Carlile, 2002), like samples, prototypes or graphs. For example, John brought a hard disk drive to a meeting with Jack, to illustrate the problems he was facing with another storage technology: ‘most technologists are familiar with hard disk technology. Because of the similarities, I use it to explain the issues I am working on’ (E98). Yet, problem owners were reluctant to provide their own analysis of the problem because that would not have been fully comprehensible for someone engaged in other practices and might also narrow a contributor’s focus.

Connecting usually occurred iteratively. Potential contributors frequently asked questions to get more information on the problem. These included questions of clarification [‘What is the scale unit at the left axis?’ (E153)] and more penetrating questions [‘Does gravity have any effect here?’ (E40)]. This latter question opened a new problem for Jason because he had not considered that possibility and subsequent analysis showed him that gravity indeed had a significant effect. The interactive and iterative exploration of a problem helps to see new dimensions of that problem and to ask questions that the problem owner was not aware of in advance (cf. Miyake and Norman, 1979). The activity of connecting did not have a clear ending. Frequently, ideas about possible solutions prompted problem owners to explain more about the context of their problem, for example, to explain why a particular solution would not be feasible.

82 H. Berends et al.

None of the episodes of thinking along aimed at the creation of shared knowledge, as these interactions did not integrate or reconcile perspectives, but made one perspective productive for a problem embedded in other practices. Nevertheless, a basic level of shared understanding facilitated thinking along. Returning to the initial example: although Richard and Andrew were engaged in different domains, they shared the experience of doing research, worked in the same industrial laboratory and shared general natural science concepts. The notion of viscosity, for example, played a role in the practices of both Richard and Andrew. Such a minimal level of shared practice helped to make sense of a problem in another domain.

5.3 Contributing

The activity of contributing includes those moves in which suggestions are brought forward and evaluated. Contributors suggested technical solutions, hypotheses, potential experiments and drew conclusions. Characteristic for thinking along is that hypotheses and solutions are completely new or elaborations of existing ideas. For example, in E209 (see Table 2), Malcolm described three questions he had with regard to the improvement of a new type of chemical process in a lunch bag seminar. This process was new to most of the attendants, but after some probing into the problem, they came up with several suggestions on how to improve the process. In E41, Jason presented a specific coating problem he had been working on. Another group member, Chris, suggested: ‘can’t you dry under vacuum?’ Of course, doing something in vacuum is not new, but as a possible solution in this case, it was new for both Chris and Jason.

Problem owners also introduced ideas. In E37 (see Table 2), Marc explained a coating defect: ‘I don’t understand it, for it is the same liquid as I am normally using and then there is no problem. Only now I am using it in black’. A couple of seconds later, he goes on: ‘but maybe … the pattern is there all the time, but you don’t see it. I might be seeing it now because of the dark colour’. Explaining an issue to someone else may suddenly result in new insight – sometimes called self-explanation (Chi et al., 1994). Further, ideas from contributors also triggered new ideas on the side of the problem owner.

Both actors also engaged in the evaluation of ideas, evidenced by moves like agreeing, rejecting, giving arguments and questioning. Due to the lack of a shared perspective, there was limited common ground to evaluate ideas. Problem owners and contributors frequently held different evaluation criteria. In the second part of episode E37 (see Table 2), Marc, a chemist, provided Jason, a physicist, with the suggestion to add a particular anti-cratering agent to solve a spin coating problem. However, Jason remarked afterwards: ‘he doesn’t know why, but it works. That’s typical for a chemist. It is the eternal difference: physicists try to understand a system and keep it simple. Chemists say ‘just add something’’. In his physicist way of working, the solution provided by Marc was no option for Jason.

Moreover, problem owners were often not fully able to evaluate the justification provided by contributors because they were unfamiliar with its background. If possible, suggestions were put to the test to see if they ‘worked’. For example, Richard tested Andrew’s hypothesis by handling the bottles of coating liquid differently. He found out that his problem did not return after he had changed his procedure and so he incorporated the alternative procedure in his practices. In the process of thinking along, both

Thinking along 83

suggestions and evaluations often prompted problem owners to iterate and explain more details on their problems.

5.4 Concluding

Thinking along was found to be an open-ended process. If successful, problem owners got fresh insights and possible solutions regarding the issues they were working on. It was up to them to benefit from it. The researchers who thought along had little say in how their suggestions were used because they were not involved in the project. Although outsiders could suggest that an idea be considered or reconsidered, they remained outsiders and their ideas could always be contested and dropped. These interactions did not end with a joint decision, neither were they aimed at consensus, thereby, avoiding the issues associated with striving for agreement across knowledge domains (cf. Boland and Tenkasi, 1995; Carlile, 2004).

For example, at the end of a meeting between John – a physicist investigating forces acting on dust particles – and Jack – specialised in designing filters and other dust management techniques, Jack asked: ‘was all of this of any use to you?’ (E98). Actually, when John reflected on some of the ideas brought forward by Jack, he commented: ‘that’s not relevant at all. He kept emphasising the issue of leakage free assembly. And I thought ‘come on, skip that’. He was just presenting possible solutions, but that wasn’t the issue for me’. In a post-hoc discussion of this episode, Jack remarked: ‘John is the one who should assess it. I am just an advisor on the sidelines. One should never try to force anything upon another researcher. If the other thinks that he knows best, who am I to say that he doesn’t?’

Not all thinking along episodes resulted in clear benefits. Usually, that was without consequence because future collaboration did not depend on it. While thinking along resulted in several significant breakthroughs, sometimes, contributors failed to come up with any useful idea. The episode between Jack and John, alluded to above, remained somewhat unsatisfactory for John, the problem owner. Yet, there was no need to continue until the interaction would deliver some particular result. Nevertheless, even episodes that were not directly beneficial helped to increase meta-knowledge and the approachability of others. As Richard gratefully remarked after another episode: ‘I went for something completely different. But they are so clever. I will ask him questions again’.

6 Discussion

Our findings contribute to a more detailed understanding of knowledge processes in an R&D environment. In particular, we offer ‘thinking along’ as a process to tap into knowledge that lies distributed across organisations. Thinking along is:

a an interactive process between two or more participants

b focused on one person’s specific problem

c in which another person helps to solve that problem

d without the contributor and problem owner having to get involved in each others’ ways of knowing.

84 H. Berends et al.

Thinking along is realised through the activities of initiating, connecting, contributing and concluding, which are executed iteratively.

We have parceled out the various element of the process so as to show how thinking along is different from knowledge transfer and transformation. Primarily, participants who think along do not need to get deeply involved in each others’ ways of knowing. In contrast, participants need to negotiate the value of different perspectives and reconcile them in the case of knowledge transformation (Carlile, 2004) and to share a common background for effective knowledge transfer to occur (Brown and Duguid, 2001). Additional differences may be seen if we look at the outcomes of the process. Unlike the case of transformation and transfer, for new ideas to emerge, thinking along does not necessarily result in common knowledge or mutual agreement.

There are facets of thinking along that are common with knowledge transformation. First, unlike transfer, thinking along and knowledge transformation processes are interactive, requiring an active role of both parties. Second, thinking along and knowledge transformation processes are focused on the reuse of knowledge for the resolution of specific innovative problems. Unlike transfer, these processes involve the adaptation or application of knowledge to a specific situation – not the replication of pre-existing knowledge (Carlile and Rebentisch, 2003; Majchrzak et al., 2004).

There is a major advantage of thinking along as a process. Thinking along enables someone to benefit from the insights that are generated when someone else with a different knowledge base bounces off ideas. Because specialised knowledge is applied by contributors, but not transferred or shared through transformation, thinking along exploits and maintains the specialisation present among organisation members. Specialists engaged in different practices with different perspectives do not become engaged in each other’s practices, nor do they adopt the other’s perspective. It would have taken Richard days if not weeks to learn all that he had to know about polymers to possibly arrive at the relevant conclusion himself. As it was, Andrew’s application of that knowledge and the subsequent communication of his comments lasted for less than 15 minutes. Indeed, thinking along avoids much of the difficulties of knowledge transfer and transformations across boundaries (Brown and Duguid, 2001; Bechky, 2003; Carlile, 2002), rendering thinking along a far less time consuming process. Moreover, thinking along strengthens the functioning of an organisation as a distributed knowledge system (Tsoukas, 1996). This gets increasingly important since R&D organisations have to cope with fast growing and differentiating technological knowledge domains (Granstrand et al., 1997; Pavitt, 1998). Yet, thinking along differs from traditional task specialisation because any contributor is only temporarily involved with a problem and bears no responsibility for the completion of that specific task.

However, these characteristics may also be a drawback. With thinking along, little co-learning is involved. The next time Richard faces another problem with a polymer liquid, he will again need to consult Andrew or another colleague. Had he acquired the necessary expertise, he could potentially have been able to solve similar problems himself in the future.

Thinking along and knowledge transfer are not mutually exclusive. While we focused on the distinctions between these two processes, they may actually co-exist and co-contribute, even in a single episode.2 Organisation members may transfer knowledge that is equally understood due to a shared background and apply thinking along for ‘sticky’ elements of knowledge. This ‘stickiness’ may be due to epistemic boundaries (Brown and Duguid, 2001), and, relatedly, to the tacitness of knowledge (Polanyi, 1958).

Thinking along 85

The observation of some thinking along episodes in which no clear epistemic boundaries were crossed may be explained by the use of thinking along to tap into the tacit elements of knowledge.

Finally, thinking along is not a panacea for all boundary problems. Thinking along is especially helpful when participants’ tasks are not interdependent in a joint project. The problems that were addressed in thinking along were not shared problems and the participants were engaged in different projects. Consequently, they had no strong need to negotiate the value of suggested solutions. In contrast, boundary issues caused by the interdependence of actors in joint projects may require them to create mutual understanding and reconcile differences through a knowledge transformation process (Bechky, 2003; Carlile, 2004).

Our findings suggest that knowledge management practitioners should differentiate between knowledge transfer, transformation and thinking along. It is likely that they should partly shift focus from knowledge transfer to the other two strategies because knowledge transfer is still dominant in knowledge management (cf. Chang et al., 2008; Soosay and Hyland, 2008). In line with the personalisation approach to knowledge management (Hansen et al., 1999), managers can create systematic opportunities for thinking along (e.g., emergent venues; institutionalising boundary spanning opportunities) and support the development of meta-knowledge on who knows what and who works on what problems [i.e., an organisational level transactive memory (Anand et al., 1998)].

7 Conclusions

Members of innovative organisations frequently face novel, unexpected problems that cannot be solved from their own perspective alone (Van Looy and Debackere, 2001). Specialisation that helps people to create their own perspective ends up creating boundaries that are characterised by a lack of common knowledge (Carlile, 2002). In this paper, we described thinking along, a process that reconciles the need for specialisation and the need to tap into the diversity of knowledge. As people think along, contributors are more than ‘senders’ and problem owners are more than just ‘receivers’ (cf. Shannon and Weaver, 1949). In thinking along, a problem owner and a contributor are both engaged in connecting and contributing to a problem from across knowledge boundaries. By thinking along, actors are enabled to draw upon the explicit knowledge and tacit knowing of specialists without the burden of having to become involved with the other persons’ practice or having to reach mutual agreement. Thinking along, thus, enables a broad utilisation of knowledge while at the same time retaining specialisation. As such, it presents a perspective on knowledge management in R&D that is better able to deal with the ongoing trend towards distributed knowledge systems.

We have documented the process of thinking along from our field observations in two research settings. To further understand it, we need to explore this process in other settings as well. What are the types of problems that may benefit from thinking along and what are the challenges involved? To what extent might we find similar processes to hold in other research organisations and in other settings? How might organisations enhance their capacity for thinking along? Relatedly, how might thinking along be encouraged to complement transfer and transformation where such processes are required? These and other related questions provide useful lines for future inquiry.

86 H. Berends et al.

References Akgün, A.E., Byrne, J.C., Keskin, H. and Lynn, G.S. (2006) ‘Transactive memory system in new

product development teams’, IEEE Transactions on Engineering Management, Vol. 53, No. 1, pp.95–111.

Allen, T.J. (1977) Managing the Flow of Technology, MIT Press, Cambridge. Amin, A. and Roberts, J. (2008) ‘Knowing in action: beyond communities of practice’, Research

Policy, Vol. 37, No. 2, pp.353–369. Anand, V., Manz, C. and Glick, W. (1998) ‘An organizational memory approach to information

management’, Academy of Management Review, Vol. 23, No. 4, pp.796–809. Andrews, K.M. and Delahaye, B.L. (2000) ‘Influences on knowledge processes in organizational

learning’, Journal of Management Studies, Vol. 37, No. 6, pp.797–810. Argote, L. (1999) Organizational Learning: Creating, Retaining and Transferring Knowledge,

Kluwer, Dordrecht. Argote, L. and Ingram, P. (2000) ‘Knowledge transfer: a basis for competitive advantage in firms’,

Organizational Behavior and Human Decision Processes, Vol. 82, No. 1, pp.150–169. Ashby, W.R. (1956) An Introduction to Cybernetics, Chapman and Hall, London. Bechky, B.A. (2003) ‘Sharing meaning across occupational communities: the transformation of

understanding on a production floor’, Organization Science, Vol. 14, No. 3, pp.312–330. Boland, R.J. and Tenkasi, R.V. (1995) ‘Perspective making and perspective taking in communities

of knowing’, Organization Science, Vol. 6, No. 4, pp.1021–1046. Brandon, D.P. and Hollingshead, A.B. (2004) ‘Transactive memory systems in organizations:

matching tasks, expertise and people’, Organization Science, Vol. 15, No. 6, pp.633–644. Brown, J.S. and Duguid, P. (1991) ‘Organizational learning and communities of practice: toward

a unified view of working, learning and innovation’, Organization Science, Vol. 2, No. 1, pp.40–57.

Brown, J.S. and Duguid, P. (2001) ‘Knowledge and organization: a social-practice perspective’, Organization Science, Vol. 12, No. 2, pp.198–213.

Carlile, P.R. (2002) ‘A pragmatic view of knowledge and boundaries: boundary objects in new product development’, Organization Science, Vol. 13, No. 4, pp.442–455.

Carlile, P.R. (2004) ‘Transferring, translating and transforming: an integrative framework for managing knowledge across boundaries’, Organization Science, Vol. 15, No. 5, pp.555–568.

Carlile, P.R. and Rebentisch, E.S. (2003) ‘Into the black box: the knowledge transformation cycle’, Management Science, Vol. 49, No. 9, pp.1180–1195.

Chang, Y.H., Chen, J.W. and Lin, B. (2008) ‘KMsharer: an information technology approach to enable knowledge management services’, International Journal of Technology Management, Vol. 43, Nos. 1–3, pp.252–265.

Chi, M.T.H., De Leeuw, N., Chiu, M.H. and Lavancher, C. (1994) ‘Eliciting self-explanations improves understanding’, Cognitive Science, Vol. 18, No. 3, pp.439–477.

Constant, D., Sproull, L. and Kiesler, S. (1996) ‘The kindness of strangers: on the usefulness of weak ties for technical advice’, Organization Science, Vol. 7, No. 2, pp.119–135.

Demsetz, H. (1988) ‘The theory of the firm revisited’, Journal of Law, Economics, and Organization, Vol. 4, No. 1, pp.141–161.

Dougherty, D. (1992) ‘Interpretative barriers to successful product innovation in large firms’, Organization Science, Vol. 3, No. 2, pp.179–202.

Frederiksen, L.F., Hemlin, S. and Husted, K. (2004) ‘The role of knowledge management in R&D: a survey of Danish R&D leaders’ perceptions and beliefs’, International Journal of Technology Management, Vol. 28, Nos. 7–8, pp.820–839.

Galunic, D.C. and Rodan, S. (1998) ‘Resource combinations in the firm: knowledge structures and the potential for Schumpeterian innovation’, Strategic Management Journal, Vol. 19, No. 12, pp.1193–1201.

Thinking along 87

Garud, R. and Kumaraswamy, A. (2005) ‘Vicious and virtuous circles in the management of knowledge: the case of Infosys technologies’, MIS Quarterly, Vol. 29, No. 1, pp.9–33.

Garud, R. and Nayyar, P. (1994) ‘Transformative capacity: continual structuring by inter-temporal technology transfer’, Strategic Management Journal, Vol. 15, No. 5, pp.365–385.

Gherardi, S. (2006) Organizational Knowledge: The Texture of Workplace Learning, Blackwell, Oxford.

Goffman, E. (1981) Forms of Talk, Basil Blackwell, Oxford. Granstrand, O., Patel, P. and Pavitt, K. (1997) ‘Multi-technology corporations: why they have

‘distributed’ rather than ‘distinctive core’ competences’, California Management Review, Vol. 39, No. 4, pp.8–25.

Grant, R.M. (1996a) ‘Prospering in dynamically-competitive environments: organizational capability as knowledge integration’, Organization Science, Vol. 7, No. 4, pp.375–387.

Grant, R.M. (1996b) ‘Toward a knowledge-based theory of the firm’, Strategic Management Journal, Vol. 17, Winter Special Issue, pp.109–122.

Grant, R.M. (2001) ‘Knowledge and organization’, in I. Nonaka and D.J. Teece (Eds.): Managing Industrial Knowledge, Sage, London.

Hansen, M.T., Nohria, N. and Tierney, T. (1999) ‘What’s your strategy for managing knowledge?’, Harvard Business Review, Vol. 77, No. 2, pp.106–116.

Hutchins, E. (1995) Cognition in the Wild, MIT Press, Cambridge, MA. Knorr-Cetina, K. (1999) Epistemic Cultures: How the Sciences Make Knowledge, Harvard

University Press, Cambridge, MA. Latour, B. (1987) Science in Action, Harvard University Press, Cambridge, MA. Lave, J. and Wenger, E. (1991) Situated Learning, Cambridge University Press, Cambridge. Lee, F. (1997) ‘When the going gets tough, do the tough ask for help? Help seeking and power

motivation in organizations’, Organizational Behavior and Human Decision Processes, Vol. 72, No. 3, pp.336–363.

Leonard-Barton, D. (1995) Wellsprings of Knowledge: Building and Sustaining the Sources of Innovation, Harvard Business School Press, Boston.

Levin, D.Z. and Cross, R. (2004) ‘The strength of weak ties you can trust: the mediating role of trust in effective knowledge transfer’, Management Science, Vol. 50, No. 11, pp.1477–1490.

Lincoln, Y.S. and Guba, E.G. (1985) Naturalistic Inquiry, Sage, London. Majchrzak, A., Cooper, L.P. and Neece, O.E. (2004) ‘Knowledge reuse for innovation’,

Management Science, Vol. 50, No. 2, pp.174–188. Miles, M.B. and Huberman, M.A. (1994) Qualitative Data Analysis, 2nd ed., Sage, London. Miller, D.J., Fern, M.J. and Cardinal, L.B. (2007) ‘The use of knowledge for technological

innovation within diversified firms’, Academy of Management Journal, Vol. 50, No. 2, pp.308–326.

Miyake, N. and Norman, D.A. (1979) ‘To ask a question, one must know enough to know what is not known’, Journal of Verbal Learning and Verbal Behavior, Vol. 18, No. 3, pp.357–364.

Nicolini, D., Gherardi, S. and Yanow, D. (Eds.) (2003) ‘Introduction: toward a practice-based view of knowing and learning in organizations’, Knowing in Organizations: A Practice-based Approach, pp.3–31, ME Sharpe, New York.

Orr, J. (1990) ‘Sharing knowledge, celebrating identity: war stories and community memory in a service culture’, in D.S. Middleton and D. Edwards (Eds.): Collective Remembering, pp.169–189, Sage Publications, Beverly Hills.

Pavitt, K. (1998) ‘Technologies, products and organization in the innovating firm: what Adam Smith tells us and Joseph Schumpeter doesn’t’, Industrial and Corporate Change, Vol. 7, No. 3, pp.433–452.

Pentland, B.T. (1992) ‘Organizing moves in software support hot lines’, Administrative Science Quarterly, Vol. 37, No. 4, pp.527–548.

88 H. Berends et al.

Polanyi, M. (1958) Personal Knowledge, Routledge and Kegan Paul, London. Postrel, S. (2002) ‘Islands of shared knowledge: specialization and mutual understanding in

problem solving teams’, Organization Science, Vol. 13, No. 3, pp.303–320. Ravasi, D. and Verona, G. (2001) ‘Organising the process of knowledge integration: the benefits of

structural ambiguity’, Scandinavian Journal of Management, Vol. 17, No. 1, pp.41–66. Rui, M.J., Yang, J., Hutchinson, J. and Wang, J.J. (2008) ‘Managing knowledge for new

product performance in the high technology industry’, International Journal of Technology Management, Vol. 41, Nos. 1–2, pp.96–108.

Sanchez, R. and Mahoney, J.T. (1996) ‘Modularity, flexibility and knowledge management in product and organization design’, Strategic Management Journal, Vol. 17, Winter Special Issue, pp.63–76.

Searle, J.R. (1969) Speech Acts, Cambridge University Press, Cambridge. Shannon, C.E. and Weaver, W. (1949) The Mathematical Theory of Communication, University of

Illinois Press, Urbana. Shih, M-H., Tsai, H-T., Wu, C-C. and Lu, C-H. (2006) ‘A holistic knowledge sharing framework

in high-tech firms: game and co-opetition perspectives’, International Journal of Technology Management, Vol. 36, No. 4, pp.354–367.

Soosay, C. and Hyland, P. (2008) ‘Managing knowledge transfer as a strategic approach to competitive advantage’, International Journal of Technology Management, Vol. 42, Nos. 1–2, pp.143–157.

Spradley, J.P. (1980) Participant Observation, Holt, Rinehart and Winston, New York. Strauss, A. and Corbin, J. (1990) Basics of Qualitative Research, Sage Publications, Newbury Park. Szulanski, G. (2000) ‘The process of knowledge transfer: a diachronic analysis of stickiness’,

Organizational Behavior and Human Decision Processes, Vol. 82, No. 1, pp.9–27. Tsoukas, H. (1996) ‘The firm as a distributed knowledge system’, Strategic Management Journal,

Vol. 17, Winter Special Issue, pp.11–25. Tyre, M.J. and Von Hippel, E. (1997) ‘The situated nature of adaptive learning in organizations’,

Organization Science, Vol. 8, No. 1, pp.71–83. Van Looy, B. and Debackere, K. (2001) ‘Innovation as a community-spanning process: looking for

interaction strategies to handle path-dependency’, in R. Garud and P. Karnoe (Eds.): Path Dependence and Creation, Lawrence Erlbaum, London.

Wenger, E. (1998) Communities of Practice, Cambridge University Press, Cambridge.

Notes 1 ‘NatLab’ is the common abbreviation of Philips Natuurkundig Laboratorium (Philips Physics

Laboratory). 2 We would like to thank an anonymous reviewer for pointing this out.