knowledge sharing mechanisms in industrial research

11
Knowledge sharing mechanisms in industrial research Hans Berends 1 , Hans van der Bij 1 , Koenraad Debackere 2 and Mathieu Weggeman 1 1 Eindhoven Centre for Innovation Studies, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands. [email protected] 2 Managerial Economics and Strategy Group, Katholieke Univesiteit Leuven, Leuven, Belgium. Previous research has firmly established the importance of knowledge sharing in Research and Development (R&D) settings. However, current theories provide only fragmented insights into the origination of knowledge sharing, and thus offer limited guidance for knowledge manage- ment practices in R&D. To integrate and extend these fragmented insights, we undertook two exploratory field studies of knowledge sharing in industrial research organizations. The contributions of this study are the following. First, we introduce three dimensions that differentiate origination mechanisms for knowledge sharing. Second, we show that some of these mechanisms correspond to mechanisms assumed in particular streams of literature, whereas others have been neglected till now. Third, based on our field studies, we show that each of these knowledge-sharing mechanisms have a different value for industrial research practices. Therefore, knowledge management in R&D should facilitate and stimulate a broad portfolio of knowledge-sharing mechanisms. 1. Introduction R esearch and development (R&D) of new technologies, products and processes require an enormous amount of knowledge. The devel- opment of a new display technology, for instance, may require knowledge of physics, mechanical engineering, chemical engineering, electrical en- gineering, information technology and marketing. Given the limitations of human cognition, it is impossible for any one individual to be an expert in all these fields. Even within one field, it is unlikely that one individual can keep abreast with all the new developments. Therefore, indivi- duals specialize in specific fields and subfields of knowledge. Because of this specialization, orga- nizations can be considered as distributed knowl- edge systems, in which knowledge is dispersed across members of the organization. A major ad- vantage of distributed knowledge systems is that they contain much more knowledge within their boundaries than systems in which all members have the same knowledge. It enables a system to carry out a wide range of tasks, like the complex set of tasks required to develop a new technology. This perspective on organizations as distributed knowledge systems is advanced by the knowledge- based theory of the Firm (Kogut and Zander, 1992, 1996; Grant, 1996a, b; Galunic and Rodan, 1998; Okhuysen and Eisenhardt, 2002), studies of distributed cognition (Hutchins, 1995; Tsoukas, 1996; Madhavan and Grover, 1998) and studies of transactive memory (Wegner, 1987; Hollings- head, 1998; Moreland, 1999; Austin, 2003). Although the specialization of research and development (R&D) staff members has its bene- fits for the development of complex products, the dispersed state of knowledge also creates a need for knowledge sharing. We refer to knowledge sharing as the deployment of knowledge in com- R&D Management 36, 1, 2006. r Blackwell Publishing Ltd, 2006. Published by Blackwell Publishing Ltd, 85 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

Upload: independent

Post on 06-Feb-2023

1 views

Category:

Documents


0 download

TRANSCRIPT

Knowledge sharing mechanisms inindustrial research

Hans Berends1, Hans van der Bij1, KoenraadDebackere

2and Mathieu Weggeman

1

1Eindhoven Centre for Innovation Studies, Eindhoven University of Technology, PO Box513, 5600 MB Eindhoven, The Netherlands. [email protected] Economics and Strategy Group, Katholieke Univesiteit Leuven, Leuven, Belgium.

Previous research has firmly established the importance of knowledge sharing in Research and

Development (R&D) settings. However, current theories provide only fragmented insights into

the origination of knowledge sharing, and thus offer limited guidance for knowledge manage-

ment practices in R&D. To integrate and extend these fragmented insights, we undertook two

exploratory field studies of knowledge sharing in industrial research organizations. The

contributions of this study are the following. First, we introduce three dimensions that

differentiate origination mechanisms for knowledge sharing. Second, we show that some of

these mechanisms correspond to mechanisms assumed in particular streams of literature,

whereas others have been neglected till now. Third, based on our field studies, we show that

each of these knowledge-sharing mechanisms have a different value for industrial research

practices. Therefore, knowledge management in R&D should facilitate and stimulate a broad

portfolio of knowledge-sharing mechanisms.

1. Introduction

Research and development (R&D) of newtechnologies, products and processes require

an enormous amount of knowledge. The devel-opment of a new display technology, for instance,may require knowledge of physics, mechanicalengineering, chemical engineering, electrical en-gineering, information technology and marketing.Given the limitations of human cognition, it isimpossible for any one individual to be an expertin all these fields. Even within one field, it isunlikely that one individual can keep abreastwith all the new developments. Therefore, indivi-duals specialize in specific fields and subfields ofknowledge. Because of this specialization, orga-nizations can be considered as distributed knowl-edge systems, in which knowledge is dispersedacross members of the organization. A major ad-vantage of distributed knowledge systems is that

they contain much more knowledge within theirboundaries than systems in which all membershave the same knowledge. It enables a system tocarry out a wide range of tasks, like the complexset of tasks required to develop a new technology.This perspective on organizations as distributedknowledge systems is advanced by the knowledge-based theory of the Firm (Kogut and Zander,1992, 1996; Grant, 1996a, b; Galunic and Rodan,1998; Okhuysen and Eisenhardt, 2002), studies ofdistributed cognition (Hutchins, 1995; Tsoukas,1996; Madhavan and Grover, 1998) and studiesof transactive memory (Wegner, 1987; Hollings-head, 1998; Moreland, 1999; Austin, 2003).

Although the specialization of research anddevelopment (R&D) staff members has its bene-fits for the development of complex products, thedispersed state of knowledge also creates a needfor knowledge sharing. We refer to knowledgesharing as the deployment of knowledge in com-

R&D Management 36, 1, 2006. r Blackwell Publishing Ltd, 2006. Published by Blackwell Publishing Ltd, 859600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

munication with others. Members of an R&Dorganization need to share knowledge to create acommon understanding of the problems at hand,and to coordinate activities (Katz and Allen,1982; Hoopes and Postrel, 1999). Furthermore,R&D project members will often not find all theappropriate knowledge within their group andtherefore they are obliged to import informationand ideas from outside their group or project(Tushman, 1978). Also, innovations often arisefrom the re-combination of pieces of knowledgethat may be present at different places in the orga-nization (Galunic and Rodan, 1998). These andother reasons explain the positive contribution oftechnical communication and knowledge sharingto the performance of R&D activities found re-peatedly in decades of research (e.g. Pelz andAndrews, 1966; Rosenbloom and Wolek, 1970;Allen, 1977; Tushman, 1978; Keller, 1994). Giventhe acceleration of the creation of technical know-ledge and the increasing strategic importance ofknowledge, the importance of knowledge sharingmay even be increasing (Badaracco, 1991; Non-aka, 1994). Therefore, the field of knowledgemanagement has set out to improve knowledgesharing within organizations, in general (e.g.Davenport and Prusak, 1998), and within R&D,in particular (Kerssens-van Drongelen et al.,1996; Miller and Morris, 1999; Collinson, 2001).

An important aspect of knowledge sharing is itsorigination. We define the origination of knowl-edge sharing as the way in which an instance ofknowledge sharing is brought about. Earlierauthors have identified the problem of the origi-nation of knowledge sharing as the problem ofconnecting those who need knowledge with thosewho have it (Huber, 1991; Gray and Meister,2004), of connecting problems and solutions(Hargadon and Sutton, 1997) and of detectingopportunities for knowledge sharing (Galunicand Rodan, 1998). The origination of knowledgesharing is far from trivial. Organization memberscannot oversee all opportunities for useful knowl-edge sharing (Huber, 1991, p. 107; Galunic andRodan, 1998). They are confronted with ‘radicaluncertainty’ (Tsoukas, 1996), which means thatthey often do not know what extra knowledgemight be relevant for them or for others.

Existing theory provides little systematic in-sight into the different ways in which knowledgesharing can originate, and therefore it offerslimited guidance for the practice of knowledgemanagement. Different streams in the literatureassume specific, yet contrasting knowledge-shar-ing mechanisms. First, a large share of the litera-

ture assumes that knowledge sharing is initiatedby someone searching for a specific piece ofknowledge and retrieving it from someone elsewho has it. In many empirical studies on thesubject of technological knowledge transfer (Szu-lanski, 1996; Hansen, 1999; Hoopes and Postrel,1999; Borgatti and Cross, 2003; Majchrzak et al.,2004), in transactive memory studies (e.g.Wegner, 1987; Hollingshead, 1998; Moreland,1999; Austin, 2003) and in the information seek-ing literature (e.g. Pinelli et al., 1993; Leckie et al.,1996; Anderson et al., 2001) this mechanism,which can be called ‘information retrieval’, isassumed. A second mechanism that is assumedin some bodies of literature is the contribution ofinformation to a group discussion on one’s owninitiative. This mechanism is presupposed in theliterature on information pooling (Stasser andTitus, 1987; Stasser et al., 1995; Okhuysen andEisenhardt, 2002) and suggestion systems (VanDijk and Van den Ende, 2002). Third, the domi-nant knowledge management literature focuseson the realization of knowledge transfer by col-lecting knowledge and making it available at acentral place (e.g. Hansen and Haas, 2001). Fi-nally, the literature on collaborative problemsolving (e.g. Okada and Simon, 1997) and brain-storming does not focus on the transfer of existinginformation or knowledge but on the creation ofnew ideas during interaction.

So, different streams of literature have workedwith contrasting models of the knowledge-sharingprocess. These differences, however, have notbeen acknowledged, nor have the suggested me-chanisms been systematically compared. Differ-ent perspectives have been pursued in relativeneglect of each other (Mohammed and Dumville,2001). Particular mechanisms may have beenoverlooked. Consequently, someone attemptingto facilitate knowledge sharing in R&D is likely tostart with too narrow a perspective.

In order to integrate and extend these existingviews, we undertook an exploratory study ofknowledge sharing in two industrial researchorganizations. The contributions of this studywere the following. First, we introduced threedimensions to differentiate between knowledge-sharing mechanisms. Second, we showed thatsome of these mechanisms corresponded to thoseassumed in particular streams of literature,whereas other mechanisms had been neglected.We also identified a number of biases in thecurrent literature. Third, based upon our fieldstudies, we demonstrated that each of theseknowledge-sharing mechanisms have different

Hans Berends, Hans van der Bij, Koenraad Debackere and Mathieu Weggeman

86 R&D Management 36, 1, 2006 r Blackwell Publishing Ltd. 2006

values for industrial research practices. Therefore,the main lesson for knowledge management inR&D is to facilitate and to stimulate a broadportfolio of knowledge-sharing mechanisms. Be-fore turning to the findings of this study, wedescribe the research methods that were used.

2. Methodology

Several ethnographic studies have encouragedresearchers to study knowledge sharing in its na-tural context (e.g. Cicourel, 1990; Orr, 1990; Laveand Wenger, 1991). Work-related communicationis an integral part of the work of researchers inindustrial laboratories (Allen, 1977; Lynch, 1985).To investigate how knowledge sharing originates,the actual practices of researchers needed to bestudied, and to avoid existing biases and assump-tions, an exploratory research approach wasconsidered appropriate. Therefore, we undertookin-depth field studies of knowledge sharing in twoindustrial research groups. Between April 1999and December 1999, communication betweenresearchers was studied in the Buijs Group, partof the NatLab, the largest laboratory of PhilipsResearch. The Buijs Group consisted of about 25research scientists and research engineers, work-ing in the fields of solid mechanics, fluid me-chanics and thermal physics. Among otherthings, their work supported the development ofoptical storage systems and display technologies.Between March 2001 and September 2001, asimilar study was carried out at Oil and GasInnovation Research (OGIR), the exploratoryresearch group of Shell Global Solutions. Themission of OGIR is to generate innovative tech-nological options for sustainable development inthe areas of energy and mobility. The members ofOGIR were, for example, concerned with re-search into biofuels and into hydrogen as anenergy carrier.

These field studies could be classified as passiveparticipant observation (Spradley, 1980). One ofthe authors temporarily shared a room withdifferent researchers, followed them to meetingsand to their laboratories, joined them for coffeeand lunch breaks and on other social occasions,but did not actively participate in their research.In the first phase of the field studies, interviewswere held with most of the group members. In thesecond phase, a number of researchers werefollowed closely for several days. Their knowl-edge sharing interactions were observed andpartly tape-recorded. Before and after the inter-

actions, the researchers were asked to clarify thesignificance that the interactions had for them. Inmany cases, we also spoke with their interlocutorsafterwards, in order to learn their point of view aswell. Some of the tape-recorded interactions werediscussed sentence by sentence with the research-ers, either by reading the transcript or listening tothe tape together. These post hoc discussionsproved to be important to understand the knowl-edge-sharing episodes. The observed interactionscomprised chance meetings in the corridor, lunchand coffee breaks, interactions between personsoccupying the same room, project meetings, groupmeetings, reports, purposeful visits, telephone callsand e-mails. Only research-related knowledgesharing was analyzed. Over 250 interactions wereobserved or documented and numbered (e.g. E26).

Field notes, transcripts of interactions andinterviews were analyzed in line with thegrounded theory approach (Glaser and Strauss,1967; Strauss and Corbin, 1990). The groundedtheory approach is a systematic way of theorybuilding. Interactions were constantly comparedand codes were developed to capture similarities,differences and relationships. The resulting cod-ing scheme was tested and improved by havingtwo coders apply it independently to a set of newinteractions and by discussing the differences. Theimproved coding scheme was used to code 102interactions for quantitative analysis. Many ofthese remaining interactions were divided intoparts, so that every part could be classifiedexclusively as a particular mechanism of knowl-edge sharing. Altogether, 227 episodes (compris-ing whole interactions and parts of interactions)were obtained: 129 from the Buijs Group and 98from OGIR. We presented our preliminary find-ings to both research groups. This resulted insuggestions for some minor improvements.

3. Characterizing the origination ofknowledge sharing

The systematic comparison of knowledge sharingepisodes and the intensive qualitative coding ses-sions yielded three dimensions that we used todistinguish between mechanisms for the origina-tion of knowledge sharing.

3.1. Dimension 1: existing content/newcontent

The first dimension is elucidated, by comparingthe following two episodes. In E1, a biweekly

Knowledge sharing mechanisms in industrial research

r Blackwell Publishing Ltd. 2006 R&D Management 36, 1, 2006 87

group meeting of the Buijs Group, Luke1 is giventhe floor to present his research. He starts off bytalking about work he completed a couple ofmonths ago. Then he reports on the problemshe is working on currently and describes some ofhis initial results. Finally, he elaborates on hisplans for the future.

Now consider E37. In this episode, Marc has aproblem with the coating of an object. During thecoating process a pattern of irregularities isformed. He tells Jason about the problem, whoin turn asks for more details. Based on this, Jasonforms a hypothesis about the cause of the pro-blem: perhaps there is water in the coating liquid.He also comes up with a solution to the problem.In this particular conversation, which only lastedabout five minutes, something else also happened.At a certain point, Marc exclaims: ‘I don’t under-stand. It’s the same liquid I normally use and thenthere’s no problem. Only now I’m using it inblack.’ A couple of seconds later he goes on: ‘Butmaybe . . . the pattern is there all the time, but youdon’t see it. I might be seeing it now because ofthe dark color.’ Marc came up with a supplemen-tary explanation for his own problem, whichexplains why he had not noticed the irregularitiesbefore.

In E1, Luke reported on his own research.What he spoke about was not new to him; hewas merely expressing his knowledge about hisown research. By contrast, in E37 Jason andMarc each came up with a new explanation. Theirexplanations of the coating irregularities did notexist before this interaction. Therefore, we coulddistinguished between knowledge sharing thatoriginates from a selection of existing informationand knowledge sharing that involves the creativedevelopment of new hypotheses, ideas, questionsor evaluations. However, except for the distinctliterature on collaborative problem solving (e.g.Okada and Simon, 1997) and the literature onbrainstorming, to date, knowledge managementliterature has treated knowledge sharing as con-sisting of the transfer of existing information orknowledge.

3.2. Dimension 2: who determines?

The second dimension that we used to distinguishbetween knowledge-sharing mechanisms con-cerned the actor who determined the content ofknowledge sharing. First, this can be the personwho is sharing his knowledge. Second, this can bethe person one is sharing his knowledge with (‘theother’). This other person can steer the selection

of information by posing a question to which thesharing person responds. Third, managementmight determine the content of knowledge shar-ing, for instance, when a manager, a group leaderor a project leader asks someone to tell somethingto somebody else. Management can also deter-mine knowledge sharing by implementing formalprocedures for guiding information flows. Think,for example, of administrative systems and man-agement information systems that prescribe whatis to be communicated.

This second dimension is partly covered by thedistinction between push approaches and pullapproaches to information sharing (e.g. Rosen-bloom and Wolek, 1970, p. 39; Langrish et al.,1972, p. 73; Schulz, 2001,p. 664). However, pre-vious research shows a strong bias towards pullapproaches (e.g. Szulanski, 1996; Hansen, 1999;Hoopes and Postrel, 1999; Borgatti and Cross,2003).

3.3. Dimension 3: orientation

The third dimension that we found is the orienta-tion of knowledge sharing. With what objective inmind is existing information selected or newinformation developed? Does one have a parti-cular problem in mind for which the sharing ofknowledge could be useful? We found four pos-sible orientations:

� orientation towards one’s own problem (thesharing person’s problem),

� orientation towards the other’s problem,� orientation towards a shared problem,� not oriented towards a particular problem.

Consider E37 again. In this episode, Jason andMarc each came up with a new explanation forMarc’s problem. Jason’s contribution was or-iented towards someone else’s problem; he wasthinking about Marc’s problem. However, Marc’swas oriented towards his own problem. He spokeabout his problem in order to elicit a reaction.Later on, he developed a new explanation withregard to his own problem. Likewise, the selectionand development of information can be orientedtowards a shared problem or not be orientedtowards a particular problem at all.

Although we do not rule out the discovery ofadditional dimensions, further use of these threedimensions confirmed that they were sufficient topinpoint knowledge-sharing mechanisms that areassumed in the literature, and to uncover ne-glected mechanisms.

Hans Berends, Hans van der Bij, Koenraad Debackere and Mathieu Weggeman

88 R&D Management 36, 1, 2006 r Blackwell Publishing Ltd. 2006

4. Mechanisms for the origination ofknowledge sharing

In the previous section, we identified three dimen-sions that could be used to characterize theorigination of knowledge sharing. In this section,we apply these dimensions to create taxonomy ofknowledge-sharing mechanisms. Three dimen-sions, with two, three and four discrete valuesrespectively, yield 24 logically possible knowl-edge-sharing mechanisms. Out of these 24 possi-ble mechanisms, 16 were found in the episodesthat we analyzed (see Table 1). Each of thesemechanisms is described in detail in Berends(2003). Here we first discuss four mechanismsthat were found frequently and that are assumedin particular streams of research. This is followedby a description of three central mechanisms thathave received little attention in the literature.

4.1. Diffusion

We speak of diffusion when members of anorganization select and communicate existing in-formation without being oriented towards a par-ticular problem. The knowledge sharing is notmeant to help anyone in particular. Nevertheless,it occurs frequently (26 out of 227 cases). Wefound several forms of diffusion. In both theOGIR and the Buijs Group, researchers wereused to write reports and publications and tohold seminars on completed research. Anotherform of diffusion was to recount success stories or

to tell about failures during lunch or other socialoccasions. An example of diffusion is E206. Tworesearchers at OGIR had installed a new piece ofmeasuring equipment that was unique in theworld. To celebrate this joyful occasion, theyinvited all the group members to drop by forcoffee and cake. During this meeting, one of themdemonstrated the apparatus to the others andproudly told them about its features and aboutthe preliminary results obtained with it.

Storing information on an intranet or in alibrary can also be considered as a form of diffu-sion, because the content of what is stored is oftendetermined by the author and is not orientedtowards a problem someone has at that moment(e.g. Hansen and Haas, 2001). Therefore, thismechanism corresponds to the model of knowl-edge sharing that is assumed in much of themainstream knowledge management literature.However, it should be acknowledged that infor-mation transfer through a repository is usuallyenabled by a search engine, giving the personreceiving the information more influence overwhat knowledge is – indirectly – transferred andmaking diffusion more like information retrieval.

4.2. Information retrieval

This mechanism is assumed and described mostoften in the literature. Someone who needs a par-ticular piece of knowledge or information obtainsit by asking someone who has it. The content ofknowledge sharing is thus not determined by the

Table 1. Frequencies of origination mechanisms and their effects .

Content Determinedby

Oriented towardsa problem of

Discussed inthis article as

Frequency Directcontributions

Indirectcontributions

Existing Sharing person None Diffusion 26 5 23Existing Sharing person The other Pushing 18 12 8Existing Sharing person Sharing person 54 12 38Existing Sharing person Both persons Information pooling 20 15 4Existing The other None 8 0 8Existing The other The other Information retrieval 16 13 6Existing The other Sharing person 7 0 7Existing The other Both persons 5 3 2Existing Management None 3 1 3Existing Management The other 1 1 0Existing Management Both 1 0 1New Sharing person None 3 0 0New Sharing person The other Thinking along 37 26 5New Sharing person Sharing person Self-suggestion 12 12 5New Sharing person Both Collaborative

problem solving14 12 0

New The other Sharing person 2 0 1Total 227 113 111

‘Sharing person’ refers to a persons who shares his knowledge; ‘the other’ refers to a person he shares his knowledge with.

Knowledge sharing mechanisms in industrial research

r Blackwell Publishing Ltd. 2006 R&D Management 36, 1, 2006 89

sharing person but by the other and is orientedtowards the other’s problem. This mechanism isassumed in empirical studies on knowledge trans-fer (e.g. Hansen, 1999; Borgatti and Cross, 2003),within transactive memory studies (e.g. Wegner,1987) and in the information-seeking literature(e.g. Leckie et al., 1996). Gray and Meister (2004)refer to it as knowledge sourcing. We found thismechanism in 16 out of 227 cases. In these cases,knowledge sharing was typically associated withthe description of research findings, materialcharacteristics, theories, technologies and litera-ture. For example, in E234 Herman tells the heattransfer coefficient of a certain material to Geof-frey, after Geoffrey asked for it.2 In short, in-formation retrieval is particularly associated withtransferring factual information.

4.3. Information pooling

In 20 cases, the person sharing information choseto do so because of a problem shared with others.We labeled this mechanism ‘information pooling’,in line with the literature in which it is assumed(e.g. Stasser and Titus, 1987). Researchers work-ing together on projects often need to poolinformation. In E102, during a meeting at theNatLab, the members of a certain project keepeach other informed on the progress of the sub-problems they are working on, report on meetingsthey have attended and about relevant ideas theyhave heard. A suggestion system (Van Dijk andVan den Ende, 2002) is an institutionalized formof information pooling. Information pooling doesnot only consist of transferring factual informa-tion, but it may also concern questions, sugges-tions and instructions.

4.4. Collaborative problem solving

This mechanism consists of developing new in-formation with regard to a shared problem. Thismechanism is associated with, for example, com-ing up with suggestions for technical solutions,new research ideas and experiments, with theconstruction of arguments and questions andwith calculating, trying or observing somethingon the spot. This evokes the image of two personsworking together at the laboratory bench or of aproject team in discussion. Typically, these pro-cesses were studied in research on brainstormingand collective problem solving (e.g. Okada andSimon, 1997; Huang and Newell, 2003).

The four knowledge-sharing mechanisms de-scribed above were found in the episodes ob-

served at OGIR and in the Buijs Group andcorrespond to models of knowledge sharing thatare assumed in particular streams in the literature.Hence our research confirmed their relevance.However, three other mechanisms that were fre-quently used in our research groups have receivedlittle attention in the mainstream literature onknowledge sharing. The in-depth grounded casestudy approach enabled us to identify and tocharacterize them. They are introduced belowand their value will be clarified in the followingsections.

4.5. Pushing

In the origination mechanism called pushing, thesharing person chooses to provide someone elsewith existing information. In this respect pushingresembles diffusion and information pooling. Butin contrast to diffusion and information pooling,pushing is oriented towards someone else’s pro-blem. Pushing involves thinking that the otherperson needs to know something, or that certaininformation might be useful for his researchactivities. Take for example E165, one of the 18episodes in which this mechanism was found. Peteoffers to show Richard various printing techni-ques that he has studied. He knew that Richardhad recently started work as a research engineeron a micro-contact printing project. Pete thoughtit might be useful for Richard to learn somethingabout existing techniques and therefore ‘pushed’his knowledge on to him. Pushing is typical ofgatekeepers (Allen and Cohen, 1969; Allen, 1977),who monitor (external) developments and pass onto their colleagues what they think might beuseful to them.

4.6. Thinking along

In 37 knowledge-sharing episodes, someone de-veloped new ideas with regard to someone else’sproblem. We called this type of interaction think-ing along. For example, in a certain episode Johnshows Peter, an expert in fracture mechanics, adisc that had broken during an experiment. Peterinvestigates the disc and draws a conclusion aboutthe cause of the fractures. This conclusion did notexist before the interaction. Likewise, thinkingalong may yield new ideas, hypotheses or ques-tions. Thinking along is not confined to informalmeetings between two researchers. Presentationsat group meetings and reviews of manuscriptsalso pose thinking-along opportunities.

Hans Berends, Hans van der Bij, Koenraad Debackere and Mathieu Weggeman

90 R&D Management 36, 1, 2006 r Blackwell Publishing Ltd. 2006

4.7. Self-suggestion

In the same way as one can think about someoneelse’s problem, one can also think about one’sown problem during interaction. The need toexplain one’s own problem or the need to defendone’s own ideas stimulates one to come up withnew explanations, solutions, arguments and con-clusions. This mechanism – self-suggestion – wasidentified 12 times. Above we described episodeE37, in which Marc came up with an alternativeexplanation for the coating problem he had. E53is a similar example. Robin is in Jason’s office.Jason wonders whether it would be possible toshow the workings of a derotator on an overheadprojector. It requires a rotating and a fixed pictureto be shown simultaneously. Jason starts drawingon the whiteboard and soon finds a simple way ofdoing it. ‘Brilliant!’ he exclaims, ‘that’s somethingwe could demonstrate at the conference. I’ll askGerald if he can build it.’ Some researchersremarked that they purposefully talked to others,neither to help them nor to obtain a usefulreaction, but to force themselves to structure theirown thoughts.

5. The Effectiveness of knowledge-sharingmechanisms

Apart from the taxonomy of knowledge-sharingmechanisms that we introduced in the previoussection, we also studied the effectiveness of thevarious knowledge-sharing mechanisms. Is onetype of mechanism more useful than another, oruseful in a distinct way? A further comparison ofepisodes with different origination mechanismsdisclosed their heterogeneous contribution to in-dustrial research practices.

A first-order distinction can be made betweendirect contributions and indirect contributions.The contribution of knowledge sharing is directwhen it helps to solve a problem that a researcheris currently working on. Indirect contributionsare those effects that are potentially useful in thefuture. Table 1 shows that origination mechan-isms differ in the degree to which they yield directand indirect contributions.

Not surprisingly, knowledge-sharing orientedtowards someone else’s problem or a problemshared by the interlocutors is more likely to yielddirect contributions to research practices. Thisholds for the selection of existing informationoriented towards a particular problem, but alsofor the development of new ideas, hypotheses or

questions with regard to a problem that a fellowresearcher might have, or a shared problem.

It is interesting to note although, that thecharacteristics of direct contributions differacross the knowledge-sharing mechanisms identi-fied. Information retrieval, the mechanism mostfrequently assumed in the literature, is actuallyvery effective in yielding factual informationrequired, such as information on material char-acteristics, technologies and activities. This fitsthe traditional interpretation of communicationas a process of uncertainty reduction (Galbraith,1973; Tushman and Nadler, 1978).

However, other mechanisms contribute directlyin ways that differ in three respects from the‘traditional’ uncertainty reduction paradigm.First, knowledge sharing that is determined bythe sharing person (e.g. pushing and thinkingalong) frequently contributes to the developmentof the other person’s knowledge even though hehad no prior question or was not uncertain aboutsomething. For example, in E153 Andrew pre-sents results of his research on the processing ofpolymer with fibers. Andrew assumes particularmaterial properties, but afterwards Frasier doubtswhether these properties still hold when the ma-terial has fibers in it. The doubts expressed byFrasier spur Andrew to investigate the materialcharacteristics, even though he considered hisresearch complete. Thus, the remarks made byFrasier helped Andrew, although he did not havea question to begin with. Second, knowledgesharing that involves the creation of new ideas(e.g. thinking along, collaborative problem sol-ving, self-suggestion) often does not yield factualinformation or knowledge in the sense of justifiedand true beliefs, but tentative ideas, quick anddirty evaluations and critical questions. This isalso clear from the above example. Frasier did notknow that Andrew’s assumption was wrong; hejust expressed his doubts. Third, knowledge shar-ing not only reduces uncertainty or ambiguity, asthe information processing approach assumes(Galbraith, 1973; Tushman and Nadler, 1978;Daft and Lengel, 1986), it may also create un-certainty and ambiguity. In this way, knowledgesharing may tempt one to take the topic further orto reflect upon an unforeseen question. Knowl-edge sharing may increase the need for reflectioninstead of reducing it, as Andrew experienced.

Many interactions that were not oriented to-wards another person’s problem or towards ashared problem did not contribute directly toresearch practices – although cases of serendipitywere observed – but they did have an indirect

Knowledge sharing mechanisms in industrial research

r Blackwell Publishing Ltd. 2006 R&D Management 36, 1, 2006 91

contribution (see Table 1). 3 Indirect contribu-tions are not immediately useful but may beuseful at a later stage. First, whatever one hearsat a group meeting for instance and which is notuseful now, may turn out to be so in the future,because one never knows what knowledge onemight need in the future (Garud and Nayyar,1994). Second, an important share of indirectcontributions consists of learning about collea-gues. By giving and attending presentations andby relating stories at lunch, people not only learnabout new results, they also learn about theproblems that others are working on and abouttheir expertise. Chris, a research engineer at theBuijs Group, said: ‘I don’t know what the generalopinion is about me. Maybe they occasionallythink ‘my goodness, he does nothing but walkaround’. But if I seem to be just chatting some-where, that chatting is purposeful: to stay in-formed about what my colleagues know!’ (NL990817). Such indirect contributions provide theconditions for directly useful knowledge sharingin the future.

6. Conditions for knowledge sharing

The previous section provided an explanation asto why each knowledge-sharing mechanism isvaluable for R&D: they contribute in distinctiveways to the work process and outcomes of R&Dstaff. The current section provides an additionalexplanation: dissimilar origination mechanismsrequire dissimilar pre-conditions. In a particularsituation, one mechanism may be feasible whileanother may not be. We illustrate this by discuss-ing the conditions for the knowledge-sharingmechanisms that were oriented towards someoneelse’s problem.

The three ways of orienting the sharing ofexisting information towards someone else’s pro-blem or towards a shared problem (i.e. by thesharing person, by the other person and bymanagement) have different enabling conditions.For information retrieval, persons in need ofknowledge need to detect the opportunity forknowledge sharing. This means they shouldhave an idea of what knowledge they lack andbe able to turn that into a question (Miyake andNorman, 1979). For example, Collins (1974) de-scribes how researchers working on the construc-tion of TEA lasers seemed cooperative byresponding to questions asked by visitors fromcompeting organizations. The visitors, however,had limited insight to what they could learn about

the TEA lasers and as a result, they did not askthe right questions. And because the researchersdid not originate knowledge sharing by them-selves they were able to keep crucial knowledgeas a secret. So, this condition for informationretrieval is not always fulfilled.

Furthermore, effective information retrievalrequires knowledge of who knows what. One ofthe researchers of the Buijs Group stated: ‘If I en-counter a problem, I first go around in my group.I’ll go to Frasier, Henry or Pete, to the person ofwhom I think ‘he is most knowledgeable about it’.You know what your colleagues do! If I want tomeasure the thickness of a layer, I go to Patrick orMitchell. Microscopy: that’s Peter. Image proces-sing: Paul. You ought to know about that!’ (NL990817). The importance of knowledge of theexistence and whereabouts of knowledge wasstressed particularly in Granovetter’s (1973) ana-lysis of weak ties and in transactive memorystudies (e.g. Wegner, 1987; Hollingshead, 1998).Wegner (1987) calls this knowledge about knowl-edge ‘meta-knowledge’. It is possible to enhanceone’s own meta-knowledge with the meta-knowl-edge of others. In the Buijs Group and at OGIR,it happened frequently that the person who wasasked a question did not have an answer, but wasable to refer the information seeker to a thirdperson. In order to increase the researchers’ baseof meta-knowledge, the NatLab had even instal-led an office called ‘Expert Consult’, specialized inhelping researchers to find the right person.

The enabling conditions for effective pushingdiffer from the conditions for information retrie-val. Thus pushing is complementary to informa-tion retrieval. Pushing, which originates by theperson who shares his knowledge, may yield ideasthat the other person is unfamiliar with or that hehas never thought about. The technologists in thestudy of Collins (1974) knew what informationwould be valuable to their visitors whereas thevisitors themselves were unaware of it. But aprerequisite for pushing is that the person intend-ing to share knowledge knows about the activitiesand the problems of a colleague. Take for exam-ple E165. Pete knows that Richard is about tostart work on a new kind of printing techniqueand that he could therefore help Richard bytelling him about the printing techniques he hasworked on himself. Without knowledge about theactivities and problems of others, pushing wouldresult in random communication.

Persons who have knowledge and persons whoneed knowledge are likely to detect differentpossibilities for useful knowledge sharing. When

Hans Berends, Hans van der Bij, Koenraad Debackere and Mathieu Weggeman

92 R&D Management 36, 1, 2006 r Blackwell Publishing Ltd. 2006

the conditions for information retrieval are ab-sent, the conditions for pushing may be present,and vice versa. In addition, in some cases manage-ment is knowledgeable enough to know how oneperson may help someone else by sharing theparticular knowledge.

The dissimilarities in the conditions for theeffective use of origination mechanisms makeseach of them suited for different situations. Dif-ferent situations require a different emphasis oneach of the mechanisms. We elaborate this pointby using the distinction between exploration andexploitation (March, 1991). Exploration, the pur-suit of new knowledge, thrives upon the re-com-bination of ideas (Hargadon and Sutton, 1997).But because of the uncertainty and ambiguityinvolved in exploration, it is often unclear whatopportunities for knowledge re-combination exist(Galunic and Rodan, 1998). It is most likely thatthose who are directly involved in potential re-combinations, the person who has the knowledgeand the person who needs the knowledge, willdetect knowledge-sharing opportunities. In anexploratory environment, management will becapable of directing knowledge sharing less fre-quently. In exploitation, the use and developmentof things already known, it is clearer what knowl-edge can be applied in a particular situation.Therefore, management is more likely to knowwhat knowledge sharing is necessary. Standar-dized procedures for knowledge sharing will bemore valuable in exploitation. Whereas we foundfew instances of knowledge-sharing directed bymanagement in this study (see Table 1), we expectto find much more of it in a more exploitativeenvironment, such as a production setting.

Exploration and exploitation also require adifferent balance between the transfer of existinginformation and the creation of new informationwithin interactions. A central requirement for thetransfer of knowledge is that the required knowl-edge actually exists. But researchers are set towork on problems for which it is assumed that nosolution exists in advance. Otherwise, researchwould not be necessary. This implies that manyquestions cannot be answered by providing exist-ing information. In this case, colleagues couldhelp by thinking along or they could engage incollaborative problem solving. Because of the spe-cific nature of working in a research environment,the development of new ideas is of irreplaceablevalue. But in an environment that is orientedmore towards the exploitation of knowledge,mechanisms that are characterized by the transferof existing information are more important.

7. Discussion and implications

The taxonomy of origination mechanisms forknowledge-sharing generated in this article hasimportant theoretical and managerial implica-tions. Most streams in the literature have limitedthemselves to the analysis of one particular me-chanism without acknowledging this explicitly.This has led to an uncoordinated study of knowl-edge-sharing, characterized by a number of biasesand the neglect of important origination mechan-isms. In future, researchers should either incorpo-rate a broad portfolio of mechanisms in theirstudies or explicate to what knowledge-sharingmechanisms they limit themselves. From a man-agerial point of view, our taxonomy can be avaluable device for the diagnosis of knowledge-sharing problems. It can be used to determinewhich mechanisms are used predominantly andwhich are neglected.

By exploring the mechanisms used in industrialresearch and by studying them simultaneously, wefound that different origination mechanisms con-tribute to R&D in heterogeneous ways and areenabled by distinct conditions. We observed thateach of the mechanisms described in this paper isvaluable in industrial research. Nevertheless, theeffectiveness of each mechanism will differ fromcontext to context. In general, we hypothesizethat for exploration the origination of knowledgesharing by the persons who have knowledge andwho need knowledge is more important, while forexploitation the direction of knowledge sharingby management will be more important. More-over, we hypothesize that for exploration thedevelopment of new information in interac-tion is more important, and that for exploitationthe transfer of existing information is more im-portant.

Given our argument that knowledge manage-ment in R&D should consider a broad portfolioof knowledge-sharing mechanisms, the codifica-tion strategy in knowledge management (Hansenet al., 1999) is not sufficient. Codifying knowledgeand collecting it in a database or intranet does notstimulate indispensable mechanisms like pushingand thinking along. In this paper we stress that, inaddition to other factors like infrastructure andtrust, knowledge about others is an importantcondition for these mechanisms. This includesboth knowledge about the knowledge of othersand knowledge about the problems that they areworking on. This knowledge about others may befacilitated through the development of commu-nities of practice (Wenger et al., 2002).

Knowledge sharing mechanisms in industrial research

r Blackwell Publishing Ltd. 2006 R&D Management 36, 1, 2006 93

Needless to say, the concepts and findingsdiscussed in this paper should be tested andelaborated in further research. It should be ex-plored whether the same origination mechanismscan be found in other organizational functions,such as engineering and marketing. Our proposi-tions with regard to the value of different mecha-nisms for exploitation and exploration also needto be tested. Finally, more work is required on thefactors that enable, stimulate and constrain theuse of particular knowledge-sharing mechanisms.

References

Allen, T.J. (1977) Managing the Flow of Technology.

Cambridge, MA: MIT Press.

Allen, T.J. and Cohen, S.I. (1969) Information flow in

research and development laboratories. Administra-

tive Science Quarterly, 14, 1, 12–19.

Anderson, C.J., Glassman, M., McAfee, R.B. and

Pinelli, T. (2001) An investigation of factors affecting

how engineers and scientists seek information. Jour-

nal of Engineering and Technology Management, 18,

2, 131–155.

Austin, J.R. (2003) Transactive memory in organiza-

tional groups: The effects of content, consensus,

specialization and accuracy on group performance.

Journal of Applied Psychology, 88, 5, 866–878.

Badaracco, J.L. (1991) The Knowledge Link. Boston,

MA: Harvard Business School Press.

Berends, H. (2003) Knowledge sharing in industrial

research. Ph.D. Thesis, Eindhoven University of

Technology.

Borgatti, S.P. and Cross, R. (2003) A relational view of

information seeking and learning in social networks.

Management Science, 49, 4, 432–445.

Cicourel, A.V. (1990) The integration of distributed

knowledge in collaborative medical diagnosis. In Gal-

legher, J., Kraut, R.E. and Egido, C. (eds), Inte-

llectual Teamwork. Hillsdale, NJ: Erlbaum. pp 221–242.

Collins, H.M. (1974) The TEA Set: tacit knowledge and

scientific networks. Science Studies, 4, 2, 165–185.

Collinson, S. (2001) Knowledge management capabil-

ities in R&D: a UK – Japan company comparison.

R&D Management, 31, 3, 335–347.

Daft, R.L. and Lengel, R.H. (1986) Organizational

information requirements, media richness and struc-

tural design. Management Science, 32, 5, 554–571.

Davenport, T.H. and Prusak, L. (1998)Working Know-

ledge. Boston, MA: Harvard Business School Press.

Galbraith, J.R. (1973) Designing Complex Organiza-

tions. Reading, MA: Addison-Wesley.

Galunic, D.C. and Rodan, S. (1998) Resource recom-

binations in the firm: knowledge structures and the

potential for Schumpeterian innovation. Strategic

Management Journal, 19, 12, 1193–1201.

Garud, R. and Nayyar, P. (1994) Transformative

capacity: continual structuring by inter-temporal

knowledge transfer. Strategic Management Journal,

15, 5, 365–385.

Glaser, B.G. and Strauss, A.L. (1967) The Discovery of

Grounded Theory. Chicago, IL: Aldine.

Granovetter, M. (1973) The strength of weak ties.

American Journal of Sociology, 78, 6, 1360–1380.

Grant, R.M. (1996a) Prospering in dynamically-

competitive environments: organizational capability

as knowledge integration. Organization Science, 7, 4,

375–387.

Grant, R.M. (1996b) Toward a knowledge-based the-

ory of the firm. Strategic Management Journal, 17,

Winter Special Issue, 109–122.

Gray, P.H. and Meister, D.B. (2004) Knowledge sourc-

ing effectiveness.Management Science, 50, 6, 821–834.

Hansen, M.T. (1999) The search-transfer problem: the

role of weak ties in sharing knowledge across orga-

nization subunits. Administrative Science Quarterly,

44, 1, 82–111.

Hansen, M.T. and Haas, M.R. (2001) Competing for

attention in knowledge markets. Administrative

Science Quarterly, 46, 1, 1–28.

Hansen, M.T., Nohria, N. and Tierney, T. (1999)

What’s your strategy for managing knowledge? Har-

vard Business Review, 77, 2, 106–116.

Hargadon, A. and Sutton, R.I. (1997) Technology bro-

kering and innovation in a product development firm.

Administrative Science Quarterly, 42, 4, 716–749.

Hollingshead, A.B. (1998) Communication, learning,

and retrieval in transactive memory systems. Journal

of Experimental Social Psychology, 34, 5, 423–442.

Hoopes, D.G. and Postrel, S. (1999) Shared knowledge,

‘‘glitches,’’ and product development performance.

Strategic Management Journal, 20, 9, 837–865.

Huang, J.C. and Newell, S. (2003) Knowledge integra-

tion processes and dynamics within the context of

cross-functional projects. International Journal of

Project Management, 21, 3, 167–176.

Huber, G.P. (1991) Organizational learning: the con-

tributing processes and the literature. Organization

Science, 2, 1, 88–115.

Hutchins, E. (1995) Cognition in the Wild. Cambridge,

MA: MIT Press.

Katz, R. and Allen, T.J. (1982) Investigating the Not-

Invented-Here (NIH) syndrome. R&D Management,

12, 1, 7–19.

Keller, R.T. (1994) Technology – information proces-

sing fit and the performance of R&D project groups.

Academy of Management Journal, 37, 1, 167–179.

Kerssens-van Drongelen, I.C., De Weerd-Nederhof,

P.C. and Fisscher, O.A.M. (1996) Describing the

issues of knowledge management in R&D: towards

a communication and analysis tool. R&D Manage-

ment, 26, 3, 213–230.

Kogut, B. and Zander, U. (1992) Knowledge of the

firm, combinative capabilities, and the replication of

technology. Organization Science, 3, 3, 383–397.

Hans Berends, Hans van der Bij, Koenraad Debackere and Mathieu Weggeman

94 R&D Management 36, 1, 2006 r Blackwell Publishing Ltd. 2006

Kogut, B. and Zander, U. (1996) What firms do?

Coordination, identity and learning. Organization

Science, 7, 5, 502–518.

Langrish, J., Gibbons, M., Evans, W.G. and Jevons,

F.R. (1972) Wealth from Knowledge: Studies of

Innovation in Industry. London: Macmillan.

Lave, J. and Wenger, E. (1991) Situated Learning.

Cambridge: Cambridge University Press.

Leckie, G.J., Pettigrew, K.E. and Sylvain, C. (1996)

Modelling the information seeking of professionals.

Library Quarterly, 66, 2, 161–193.

Lynch, M. (1985)Art and Artifact in Laboratory Science:

A Study of Shop Work and Shop Talk in a Research

Laboratory. London: Routledge & Kegan Paul.

Madhavan, R. and Grover, R. (1998) From embedded

knowledge to embodied knowledge. Journal of Mar-

keting, 62, 4, 1–12.

Majchrzak, A., Cooper, L.P. and Neece, O.E. (2004)

Knowledge reuse for innovation. Management

Science, 50, 2, 174–188.

March, J.G. (1991) Exploration and exploitation in orga-

nizational learning. Organization Science, 2, 1, 71–87.

Miller, W.L. and Morris, L. (1999) Fourth Generation

R&D: Managing Knowledge, Technology and Innova-

tion. New York, NY: John Wiley & Sons.

Miyake, N. and Norman, D.A. (1979) To ask a ques-

tion, one must know enough to know what is not

known. Journal of Verbal Learning and Verbal Beha-

vior, 18, 3, 357–364.

Mohammed, S. and Dumville, B.C. (2001) Team men-

tal models in a team knowledge framework. Journal

of Organizational Behavior, 22, 2, 89–106.

Moreland, R.L. (1999) Transactive memory: Learning

who knows what in work groups and organizations.

In Thompson, L.L., Levine, J.M. and Messick, D.M.

(eds), Shared Cognition in Organizations. Mahwah,

NJ: Lawrence Erlbaum, pp 3–31.

Nonaka, I. (1994) A dynamic theory of organizational

knowledge creation.Organization Science, 5, 1, 14–47.

Okada, T. and Simon, H.A. (1997) Collaborative dis-

covery in a scientific domain. Cognitive Science, 21, 2,

109–146.

Okhuysen, G.A. and Eisenhardt, K.M. (2002) Integrat-

ing knowledge in groups: how formal interventions

enable flexibility.Organization Science, 13, 4, 370–386.

Orr, J. (1990) Sharing knowledge, celebrating identity:

war stories and community memory in a service

culture. In Middleton, D.S. and Edwards, D. (eds),

Collective Remembering. Beverly Hills, CA: Sage

Publications, pp 169–189.

Pelz, D.C. and Andrews, F.M. (1966) Scientists in

Organizations. New York, NY: Wiley.

Pinelli, T.E., Bishop, A.P., Barclay, R.O. and Kennedy,

J.M. (1993) The information-seeking behavior of

engineers. Encyclopedia of Library and Information

Science, 52, supplement 15, 167–201.

Rosenbloom, R.S. and Wolek, F.W. (1970) Technology

and Information Transfer. Boston, MA: Harvard Uni-

versity.

Schulz, M. (2001) The uncertain relevance of newness:

organizational learning and knowledge flows. Acad-

emy of Management Journal, 44, 4, 661–681.

Spradley, J.P. (1980) Participant Observation. New

York, NY: Holt, Rinehart and Winston.

Stasser, G., Stewart, D.D. and Wittenbaum, G.M.

(1995) Expert roles and information exchange during

discussion. Journal of Experimental Social Psychol-

ogy, 31, 3, 244–265.

Stasser, G. and Titus, W. (1987) Effects of information

load and percentage of shared information on the

dissemination of unshared information during group

discussion. Journal of Personality and Social Psychol-

ogy, 53, 1, 81–93.

Strauss, A. and Corbin, J. (1990) Basics of Qualitative

Research. Newbury Park, CA: Sage Publications.

Szulanski, G. (1996) Exploring internal stickiness: im-

pediments to the transfer of best practice within a

firm. Strategic Management Journal, 17, Winter

Special Issue, 27–44.

Tsoukas, H. (1996) The firm as a distributed knowledge

system. Strategic Management Journal, 17, Winter

Special Issue, 11–25.

Tushman, M.L. (1978) Technical communication in

R&D laboratories: the impact of project work char-

acteristics. Academy of Management Journal, 21, 4,

624–645.

Tushman, M.L. and Nadler, D.A. (1978) Information

processing as an integrating concept in organiza-

tional design. Academy of Management Review, 3,

3, 613–624.

Van Dijk, C. and Van den Ende, J. (2002) Suggestion

systems: transferring employee creativity into prac-

ticable ideas. R&D Management, 32, 5, 387–395.

Wegner, D.M. (1987) Transactive memory. In Mullen,

B. and Goethals, G.R. (eds), Theories of Group Be-

havior. New York, NY: Springer Verlag. pp 185–208.

Wenger, E., McDermott, R. and Snyder, W.M. (2002)

Cultivating Communities of Practice. Boston, MA:

Harvard Business School Press.

Notes

1. Names have been changed in order to protect

anonymity.

2. Information retrieval usually requires that the in-

formation seeker first expresses his problem or

question. The presentation of a problem of a ques-

tion in order to elicit a reaction has been coded as

the sharing of existing content, determined by the

sharing person, oriented towards a problem of the

sharing person. Naturally, this ‘preparatory’ com-

munication was found frequently (see Table 1).

3. The relative frequency of direct contributions stem-

ming from diffusion would have been higher had

this study included the use of internal reports and

journals. These can be seen as cases of diffusion as

well, but libraries and databases provide more

opportunities to select from the offered information.

Knowledge sharing mechanisms in industrial research

r Blackwell Publishing Ltd. 2006 R&D Management 36, 1, 2006 95