Garud & Kumaraswamy/Vicious and Virtuous Circlesphp.scripts.psu.edu/users/r/u/rug14/12.Viscious and...

Garud & Kumaraswamy/Vicious and Virtuous Circles

MIS Quarterly Vol. 29 No. 1, pp. 9-33/March 2005 9

SPECIAL ISSUE

VICIOUS AND VIRTUOUS CIRCLES IN THEMANAGEMENT OF KNOWLEDGE: THECASE OF INFOSYS TECHNOLOGIES1

By: Raghu GarudDepartment of ManagementLeonard N. Stern School of BusinessNew York University40 West 4th StreetNew York, NY [email protected]

Arun KumaraswamyLee Kong Chian School of BusinessSingapore Management University469 Bukit Timah RoadSingapore 259756REPUBLIC OF [email protected]

Abstract

We adopt a systems perspective to explore thechallenges that organizations face in harnessingknowledge. Such a perspective draws attention tomutually causal processes that have the potentialto generate both vicious and virtuous circles.Based on a longitudinal study at Infosys Techno-

1V. Sambamurthy and Mani Subramani were theaccepting senior editors for this paper.

logies, we conclude that knowledge managementinvolves more than just the sponsorship of initia-tives at and across different organizational levels.It also involves an active process of steeringaround and out of vicious circles that will inevitablyemerge.

Keywords: Knowledge management, increasingreturns, systems dynamics, vicious circles

Introduction

Knowledge is an important organizational resource(Penrose 1995; Winter 1987). Unlike other inertorganizational resources, the application ofexisting knowledge has the potential to generatenew knowledge (Leonard 1998; Zuboff 1984). Notonly can knowledge be replenished in use (Gid-dens 1986; Schon 1983), it can also be combinedand recombined to generate new knowledge(Garud and Nayyar 1994; Grant 1996a; Hargadon2003; Kogut and Zander 1992; Okhuyzen andEisenhardt 2002). Once created, knowledge canbe articulated, shared, stored and recontextualizedto yield options for the future (Sambamurthy et al.2003). For all of these reasons, knowledge hasthe potential to be applied across time and spaceto yield increasing returns (Fortune 1991; Shapiroand Varian 1999).


10 MIS Quarterly Vol. 29 No. 1/March 2005

Harnessing knowledge for increasing returns,however, is not an easy task. Leidner (2000), forinstance, pointed out that many knowledgemanagement initiatives have yet to yield significantorganizational improvements. Others have writtenabout “knowledge management as a double edgedsword” (Schultze and Leidner 2002), the “deadliestsins of knowledge management” (Fahey andPrusak 1998) and “knowledge traps” (Soo et al.2002). Some have documented unsuccessfulknowledge management efforts, concluding thatmanaging knowledge is not easy (Nidumolu et al.2001).

These difficulties arise because knowledgeprocesses have to be managed at and across dif-ferent organizational levels (Nonaka and Takeuchi1995). At each level, there are forces at work thatcan easily stifle the generation of new knowledge(March 1991). Across levels, the coupling of dif-ferent knowledge processes can give rise to un-anticipated negative consequences (Senge 1990).

Over time, processes that yield such negativeoutcomes can degenerate into vicious circles(Masuch 1985). Vicious circles arise when mutu-ally causal processes feed back into one anotherto lock a system into a mode of operation thatyields progressively negative outcomes (Maru-yama 1963; Masuch 1985; Senge 1990; Weick1969). In contrast, virtuous circles are those thatyield increasing returns. The challenge for anorganization is to harness its knowledge processesto generate a virtuous circle of increasing returnsdespite the ever existing potential for viciouscircles to emerge.

We adopt a systems perspective (Maruyama 1963;Masuch 1985; Perrow 1984; Senge 1990; Weick1969) to gain an understanding of the micro-processes that give rise to this challenge. Such aperspective conceptualizes knowledge processesunfolding at and across different organizationallevels as a system. It also draws attention to themutually causal processes constituting the organi-zation’s knowledge system.

We apply this perspective to a longitudinal study ofknowledge initiatives at Infosys Technologies, a

company acknowledged globally for its knowledgemanagement practices. We explore how Infosysattempted to couple knowledge processes at andacross the individual, group, and collective organi-zational levels. We find that the very initiativesundertaken to harness an organization’s knowl-edge system by generating a virtuous circle ofknowledge accumulation, reuse, and renewal canjust as easily generate vicious circles. Based onthese findings, we suggest that knowledge man-agers must employ process interventions to steeran organization’s knowledge system around or outof the vicious circles that are bound to arise.

Organizing for Knowledge

Organizing is a knowledge intensive activity. Itinvolves all of the resources that an organizationpossesses: its employees and the patterns ofinteractions among them, its knowledge reposi-tories, and its rules and routines that providecohesion. In other words, knowledge manage-ment issues pervade an organization’s people,structures, systems, and processes (Govindarajanand Gupta 2001; Grant 1996b; Hutchins 1995;Subramaniam and Youndt 2004).

Much research has focused on knowledgeprocesses and techniques with the potential toyield increasing returns. Consider, for instance,Nonaka and Takeuchi’s (1995) knowledge spiral.The knowledge spiral is based on employee inter-actions which result in repeated conversions ofknowledge between its tacit and explicit forms. Assuch interactions and conversions occur, knowl-edge spirals up from the individual to the collectivelevels of the organization, thereby generating avirtuous circle.

In drawing attention to interactions at and acrossdifferent levels of an organization, the knowledgespiral sensitizes us to a need to manage knowl-edge processes within an organization as a sys-tem (Spender 1996). A system is a set of relation-ships among constituent variables, and the fate ofthe system is determined not by any single rela-tionship, but by an overall pattern. This is because


MIS Quarterly Vol. 29 No. 1/March 2005 11

system variables are coupled by mutually causalrelationships2 that have the potential to generatecomplex nonlinear dynamics (Maruyama 1963;Weick 1969). Indeed, as Nonaka and Takeuchiconcluded, “the actual process by whichorganizational knowledge creation takes place isnonlinear and interactive” and “knowledge creationis a never-ending, interactive process” (p. 225).

Senge (1990) pointed out that mutually causalprocesses, which constitute a system, have to bemaintained in a dynamic balance between forcesthat provide continuity and those that bring aboutchange (see also Jelinek and Schoonhoven 1990).Such a balance must be maintained at and acrossorganizational levels, and a failure to do so caneasily generate negative consequences. Often,these negative consequences are manifest onlyafter a time lag, thereby resulting in interventionsthat compound problems instead of mitigatingthem.

Employing a systems perspective as an interpre-tive frame, we provide a summary review of thevast and growing literature on knowledge manage-ment. In our review, we focus on opposing forcesthat arise at and across different organizationallevels (see Figure 1 for a summary). Such anapproach facilitates a deeper understanding of theprocesses that render the management of knowl-edge a rewarding yet challenging task.

Dynamics at EachOrganizational Level

Individual level dynamics. Employees play acritical role in generating and applying knowledgewithin organizations. As “men on the spot” (Hayek1945), they deal with emergent situations inmeaningful, contextualized ways without relying oninstructions from above (Markus et al. 2002;

Tsoukas 1996). In deploying available knowledgeto address emergent situations, these employeeshave the potential to generate new knowledge.Such “exploration” through “exploitation” (March1991) can happen to the extent that employeeshave the capacity to reflect-in-action. As Schon(1983, p. 68) noted,

When someone reflects-in-action, hebecomes a researcher in the practicecontext. He is not dependent on thecategories of established theory andtechnique, but constructs a new theory ofthe unique case. Because his experi-menting is a kind of action, implemen-tation is built into his inquiry.

Yet, opposing forces may drive out such reflection.Specifically, employees accumulate and refine theknowledge required to deal with their contextsthrough a process of learning-by-doing (Argote1999; Arrow 1962; Dutton and Thomas 1985).Although learning-by-doing can generate expertisein a specific area, it can also lead to a “compe-tency trap” (Levitt and March 1988). This is be-cause learning-by-doing is a path dependent pro-cess (David 1985). Consequently, in the very actof refining existing knowledge within a taken-for-granted framework, employees may forgo oppor-tunities to renew and expand their knowledge toolkit (Swidler 1986). Moreover, as habituation setsin through learning-by-doing, an employee’s verycapacity to reflect-in-action may be compromised.

In sum, learning-by-doing can be at odds withreflection-in-action. Whereas learning-by-doingrepresents single-loop learning, reflection-in-actionrepresents double-loop learning (Argyris andSchon 1978). The balance that an organizationstrikes between these two types of learning canhave an important bearing on whether or not it canharness its knowledge system to yield a virtuousknowledge circle.

Group level dynamics. A dynamic balance alsoneeds to be maintained between the continuity thatan epistemic community offers and the impetus forchange that connections across epistemic com-munities can provide. To appreciate the need for

2According to Weick (1968, p. 81) mutual causationmeans that “the amount of influence that variable Xexerts over variable Y determines the amount Y exertsover X; and the influence of Y over X then determinesthe subsequent influence of X over Y” (see alsoMaruyama 1963).



Coupling Individual-Group

Levels

Coupli

ng

Indivi

dual-

Collec

tive

Leve

lsCoupling

Group-Collective

Levels

Collective Level

Group LevelIndividual Level

Single-loop learningvs.

Double-loop learning

Community vs.

Connections

Digital options vs.

Information overload

Coupling Across Levels

Coupling Individual-Group

Levels

Coupli

ng

Indivi

dual-

Collec

tive

Leve

lsCoupling

Group-Collective

Levels

Collective Level

Group LevelIndividual Level

Single-loop learningvs.

Double-loop learning

Community vs.

Connections

Digital options vs.

Information overload

Coupling Across Levels

Figure 1. Balancing Opposing Forces Within an Organization’s Knowledge System

this balance, consider two key perspectives ongroup knowledge. A “community of practice” per-spective (Brown and Duguid 1991; Lave andWenger 1994; Orlikowski 2002; Orr 1990) drawsattention to shared identities and beliefs among acommunity of practitioners with a common“thought world” (Dougherty 1992). As Lave andWenger (1994, p. 98) pointed out,

[A] community of practice is an intrinsiccondition for the existence of knowledge,not least because it provides the interpre-tive support necessary for making senseof its heritage. Thus, participation in thecultural practice in which any knowledgeexists is an epistemological principle oflearning.

Group cognition is also constituted by the set ofconnections established between members of awork group. Within a work group, group cognitionis constituted by the strength of the ties betweenmembers with different epistemological leanings(Garud and Kotha 1994; Sandelands and Stablein1987; Weick and Roberts 1993). Because workgroup members have different epistemologies, it ispossible for the work group to “respond as a com-plete system to meet situational demands eventhough the complexity of the task is beyond thecognitive capabilities of individual team members”(Faraj and Sproull 2000, p. 1556). Such a re-sponse is possible to the extent that unproductiveconflict is minimized by carefully shaping inter-dependencies among group members with dif-ferent epistemologies (Raghuram et al. 2001).



Organizations attempt to reconcile knowledgegenerated within epistemic communities with thatgenerated by workgroups (Levina 2002). In manydynamic systems, we may observe a duality overtime, with epistemic communities driving workgroup connections and vice versa. In some in-stances, however, knowledge derived through con-nections within workgroups may diverge fromknowledge generated within epistemic communi-ties.3 How an organization addresses this diver-gence between these two bases of knowledge hasan important bearing on its ability to generate andsustain a virtuous knowledge circle.

Collective level dynamics. The mindful appli-cation of knowledge by individuals and structuralarrangements within work groups clearly shapebehavior and learning within organizations. Yet,as March and Simon (1993, p. 8) highlighted, the“retrieval of experiences preserved in an organi-zation’s files or individuals’ memories” is alsoimportant. Indeed, an organization can enhancethe benefits accruing from knowledge processesunfolding at and across various levels if arepository exists for stocking knowledge flows.

Here, the metaphor of organizations as knowledgerepositories (Walsh and Ungson 1991) comes tomind. Such a metaphor has become all the moreimportant as information technologies enable thecreation of digital assets and options (Markus2001; Miller 2002; Sambamurthy et al. 2003). Inthis regard, corporate intranets and knowledgeportals serve as digital repositories within whichcodified organizational knowledge accumulates. Itis far easier for employees to retrieve and reuseknowledge from today’s digital repositories thanfrom the memory banks of yesteryear. Such ease

of use enhances the options value of digital repos-itories (Miller 2002; Sambamurthy et al. 2003).

Despite these benefits, digital repositories cancreate information overload (Brown and Duguid2002; Davenport and Prusak 1998). It hasbecome all too easy to accumulate knowledge indigitized form. However, after a point, search andrecontextualization costs outweigh the potentialbenefits from reusing the knowledge. Categori-zation of digitized knowledge in repositories maymitigate this problem of information overload(Bowker and Star 2000); however, categorizationschemes themselves can create other problems.Specifically, as “layers of technology accrue andexpand over space and time,” these technologyinfrastructures inherit “the inertia of the installedbase of systems that have come before” (Bowkerand Star 2000, p. 33). Consequently, users’ re-quirements may remain unmet (Markus 2001),thereby reducing knowledge reuse and thepotential for a virtuous knowledge circle to emerge.

Interactive Dynamics Across Levels

Managing opposing forces at each organizationallevel is a difficult enough task (Alavi and Leidner2001). To complicate matters, as Grover andDavenport (2001, p. 8) pointed out, knowledgeprocesses are “recursive, expanding, and oftendiscontinuous. Many cycles of generation, codifi-cation, and transfer are concurrently occurring inbusinesses.” Therefore, coupling these knowledgeprocesses, which are unfolding across levels togenerate a virtuous circle, may give rise to newchallenges.

To illustrate these challenges, we consider severalinitiatives that organizations undertake to coupleknowledge processes within and across levels.For instance, consider the institutionalization oforganizational routines as a response to com-plexity faced by employees and workgroups.Organizational routines help couple differentknowledge processes unfolding at and acrossdifferent levels (Nelson and Winter 1982). Indeed,they set the decision context that shapes individual

3Such divergence seems to have occurred as theColumbia space shuttle crisis unfolded. When theColumbia shuttle took off, a piece of foam struck one ofthe wings. In an analysis of the events that unfolded, apanel of experts concluded, “allegiance to hierarchy andprocedure had replaced deference to NASA engineers’technical expertise” (Columbia Accident InvestigationBoard 2003, p. 200). These observations suggest thatknowledge from work group connections appears to haveprevailed over knowledge from the technical community.



and collective behavior. As March and Simon(1993, p. 8) observed, “actions are chosen byrecognizing a situation as being of a familiar,frequently encountered type, and matching therecognized situation to a set of rules.”

Yet, despite these benefits, routines can easilyentrap an organization into a knowledge trajectorythat is inconsistent with the demands of itschanging environment. Organizational routines forharnessing knowledge may become so inflexiblethat they become the basis, not for dynamism, butfor stagnation. At the extreme, core capabilitiesmay become core rigidities (Leonard-Barton 1992).

Or, consider the recent attempts by many organi-zations to apply technology architectures to coupleprocesses across levels. As Latour (1991) pointedout, “technology is society made durable.” That is,fragile social processes are shaped by the pre-sence of technological artifacts that enable andconstrain social interactions in productive ways.Indeed, social rules are built into new informationtechnologies, and these rules shape social pro-cesses (DeSanctis and Poole 1994; Orlikowski1992). By facilitating the emergence of communi-ties, these built-in rules can potentially alleviateproblems generated by the interplay between tacitand explicit knowledge (Bowker and Star 2000).

Yet, the generation of communities throughtechnology architectures can give rise to newproblems. Specifically, social rules built into tech-nologies can potentially overdetermine socialprocesses (Brown and Duguid 2002; Davenportand Prusak 1998; Leidner 2000). Indeed, theserules can become so internalized and taken forgranted that self-reflection gives way to mindlessconformity (Berger and Luckman 1967; Schon1983). Such mindless conformity can generateinappropriate actions, especially in complex,dynamic environments.

Finally, consider organizational initiatives toconnect different levels by creating “markets forknowledge” (Davenport and Prusak 1998). Speci-fically, firms have been institutionalizing schemesthat incentivize individuals to share and reuse

knowledge. Such incentive schemes are mecha-nisms designed to overcome the challenges ofinducing collective action from autonomousindividuals (Oliver et al. 1985; Olsen 1965;Schelling 1978).

Despite the merits of such schemes, however, amarket-for-knowledge perspective may end updestroying community dynamics critical to the freeflow of rich tacit knowledge (Gold et al. 2001).Spontaneous social interactions become trans-formed into calculative social exchanges (Fuku-yama 1995). As a result, an organization mayhave to incur higher transaction costs within sucha market-for-knowledge than within communities(Callon 1998; Garud 1994).

Knowledge Managementin Perspective

These interactive dynamics suggest that knowl-edge processes are inherently fragile (von Kroghet al. 2000). Such fragility implies that knowledgemanagers cannot just address issues at differentorganizational levels in a piece-meal manner.Rather, they need to embrace a systemic ap-proach to knowledge management, dynamicallybalancing and trading off opposing forces at andacross different organizational levels (Senge1990). To appreciate the intricacies and chal-lenges of such a role, we present a longitudinalstudy of knowledge management initiatives at acompany acknowledged globally for its knowledgemanagement practices. In describing these initia-tives and their consequences, we offer insightsinto the generation and maintenance of a virtuousknowledge circle over time. However, first, wedescribe our research site and methodology.

Research Site and Methods

This research is an outcome of our continuingassociation with Infosys Technologies, a globalsoftware services company based in India. At the



end of its fiscal year 2004, Infosys was a U.S.$1 billion company with over 23,000 employeesand globally distributed operations. Listed on theNASDAQ Stock Market and growing annually by atleast 30 percent during the past decade, Infosys isamong the companies consistently featured byBusiness Week in its annual Info Tech 100 list. Itis also among a select group of companies to havereceived both the Asian and the Global MostAdmired Knowledge Enterprise (MAKE) awards.

In exploring knowledge management (KM) prac-tices at Infosys, we employed a naturalistic modeof inquiry wherein insights are induced throughinterpretive means (Lincoln and Guba 1985). Thisinquiry mode emphasizes procedural adequacyand credibility, which we established by employingthe steps set out in Miles and Huberman’s (1984)primer on qualitative research.

Our aim was to generalize from a case to a theory,rather than from a sample to a population.Typically, this is accomplished by iterating betweendata and theory until a stage of theoretical satura-tion is reached (Glaser and Strauss 1967). Lincolnand Guba (1985) outlined a systematic process forgeneralizing from a case to a theory. This processinvolves continually cycling through the followingfour steps: (1) purposive sampling, (2) inductivedata analysis, (3) development of grounded theory,and (4) projection of next steps. Consistent withthese steps, we began our purposive samplingwithin Infosys in the summer of 2000 by inter-viewing senior executives and mid-level managers.Over the next three years, we conducted multiplerounds of interviews with employees from differentfunctions and levels. Overall, we conducted 56interviews over a period of 3 years. We inter-viewed a few key people more than once in orderto track how their perspectives evolved over time.

The interviews themselves were semi-structuredand emergent. Participants discussed issues thatthey felt were most important for knowledgemanagement and the growth of the company.Each interview, lasting between 1 hour and 1.5hours, was taped and transcribed. Interviewedemployees pointed us to documents such as

strategic reports, analysts’ reports, presentations,white papers, and employee surveys that furtherclarified knowledge management processes andoutcomes at Infosys.

An analysis of the interview data and companydocuments enabled us to develop a more focusedunderstanding of the company’s accomplishmentsand challenges in the management of its knowl-edge. As part of our analysis, we read the inter-view transcripts and then listened to the tapedinterviews to check the transcripts for accuracy.We also read all company documents to which wehad been referred by employees. We codedstatements made during the interviews into adatabase using keywords, including the source foreach statement and the type of documentaryevidence that established the validity of claimsmade in the statements. Progressively, we com-bined these statements into broader themes.

The theorizing process was emergent. As wedeveloped our database and continued to track thecompany, working hypotheses emerged. Forinstance, we concluded that knowledge manage-ment issues pervade the entire company.Accordingly, we decided that it would not suffice tostudy only one facet of knowledge management.We also realized that the outcomes of initiatives atInfosys would be manifest only over time asknowledge processes unfolded at and acrossdifferent organizational levels. Therefore, wedecided to conduct a longitudinal analysis of thecompany’s knowledge management efforts.

As we completed the first round of interviews andanalysis, we planned our next iteration. In theprocess, we perceived a need to gain a deeperunderstanding of the drivers and outcomes of thecompany’s various knowledge initiatives. To doso, we decided to forge closer associations withthe company’s KM group and employees atvarious levels. Over the next three years, we heldperiodic interviews with members of the KM groupand employees at all levels within Infosys. Wealso communicated periodically through e-mail withmembers of the KM group and a cross-section ofemployees as we sought further information or



clarification on specific initiatives. We coded theseperiodic interviews and responses to our e-mailcommunications. Again, we went about devel-oping themes and coming up with working hypoth-eses to inform our subsequent steps.

By the end of 2002, we came to the conclusionthat it was critical for us to gain an ethnographicfeel for the dynamics at play within this company.Accordingly, a member of our research team spent45 days at the company, becoming a part of theInfosys community to observe knowledge pro-cesses unfolding at various levels first-hand. Herdetailed insider’s accounts and final debriefingreport were invaluable not only in strengtheningour working hypotheses, but also in extending ourinsights.

Concurrent with these activities, we began writinga case on Infosys, placing special emphasis on itsknowledge management initiatives and processes.In April 2002, we completed a first draft of thiscase, which we sent to the company for review andclearance. Several employees offered criticalfeedback and clarification, pointing us to additionalbenefits and problems that they perceived with thecompany’s knowledge management initiatives.We made relevant additions and changes to thecase based on this feedback and sent it back tothe company for further review. After two suchiterations, the company gave its final clearance(Garud et al. 2003).

By this time, we had developed a deep under-standing of the practices and processes thatInfosys employed to harness its distributed knowl-edge. We made periodic presentations to the KMgroup at Infosys, whose members commented onour presentations and offered additional insights.These interactions were invaluable to us in devel-oping a greater appreciation of accomplishmentsInfosys had achieved and the challenges it facedin its efforts to manage knowledge as an organiza-tion-wide resource.

We were intrigued when, in April 2002, the KMgroup at Infosys decided to change the incentivescheme it had implemented to promote contri-butions to the company’s central knowledge portal.

Our discussions and subsequent analysis led us toa key insight that we develop in this paper: Thevery initiatives undertaken to initiate a virtuousknowledge circle may yield unintended conse-quences because of the mutually causal knowl-edge processes unfolding at and across differentorganizational levels.

Knowledge Management atInfosys Technologies

An IT company like ours cannot surviveif we don’t have mechanisms to reuse theknowledge that we create….”Learn once,use anywhere” is our motto. The visionis that every instance of learning withinInfosys should be available to everyemployee.

These sentiments, offered by a member of the KMgroup at Infosys, are reflective of the company’sefforts to leverage knowledge created by itsemployees for corporate advantage. The adage“learn once, use anywhere” reinforces the con-tinual learning and reflection required for knowl-edge accumulation and reuse. It also draws atten-tion to a core belief that knowledge belongs notonly to those employees who create it, but also tothe entire company.

Infosys began efforts to transform its employees’knowledge into an organization-wide resource inthe early 1990s (see Table 1 for an abbreviatedchronology of initiatives; for complete details, seeKochikar et al. 2002). In 1992, Infosys encour-aged its employees to offer written accounts oftheir on-the-job experiences on a variety of topicsranging from technology and software develop-ment to living and behaving in foreign cultures.These nuggets of experiential knowledge—calledbodies of knowledge (BOKs)—were then shared inhard copy form among all employees. This initia-tive was an early effort on the part of Infosys tocodify knowledge generated by its employees as anatural by-product of their daily work.



Table 1. Knowledge Management Initiatives at Infosys TechnologiesYear KM Initiatives

Since1980s

• Employees hired for learnability, not just for technical knowledge.

1992 • Bodies of knowledge (BOKs) initiative launched.

1996-97 • Corporate intranet (Sparsh) launched. • Technical bulletin boards, BOKs and repositories offered through Sparsh.• CMM Level 4 certification attained.

1998 • People Knowledge Map implemented on Sparsh.

1999 • CMM Level 5 certification attained.• Central KM group chartered.• Company-wide KM program launched with emphasis on web/repository based

approach.

2000-01 • Central knowledge portal (KShop) launched.• Customization tools for KShop entry pages offered; Local repositories integrated

with KShop; corporate data made available on KShop.• Knowledge currency units (KCU) incentive scheme launched to jumpstart

contributions to KShop.• Forms and project templates changed to enable knowledge extraction using

automated tools.

2002 • Modified KCU incentive scheme implemented.• Project tracking tool implemented on KShop.• KM Prime and Knowledge Champion roles instituted.• Initiative to promote story telling and accounts of war games launched.

During the next few years, this initiative mush-roomed into a full-fledged KM effort supported bytools such as e-mail, bulletin boards, andrepositories for marketing, technical, and project-related information. In 1996, Infosys createdSparsh, the corporate intranet, to make BOKs (inHTML format), bulletin boards, and localrepositories easily accessible to all employees.Soon, Sparsh became the central informationportal for Infosys.

In late 1999, Infosys initiated a formal company-wide KM program to integrate all knowledgeinitiatives. One of the first decisions made underthis initiative was to establish a central KM groupto facilitate the company-wide KM program. Asecond key decision was to create a centralknowledge portal called KShop. Consistent withits philosophy emphasizing central facilitation of

distributed knowledge processes, the KM groupcreated a technology infrastructure, but encour-aged different practice communities within thecompany to maintain the content on KShop. Awhite paper published by the KM group (Kochikar2001) clarified this philosophy which still drives KMat Infosys:

A key success factor is to achieve theright balance between centralization anddecentralization in KM initiatives. Cen-tralization allows a greater ability toachieve organizational synergies andscale economies, but may be difficultfrom an implementation perspective. Itmay be easier to create smaller pocketsto start with. Also, ownership and individ-ual participation tends to be low asinitiatives scale up. Niche groups within



the organization may find that theirrelative cohesion facilitates such sharingbetter. Home pages, specific knowledgedatabases and utilities are best main-tained at personal/group levels, whileknowledge directories and bodies ofknowledge are better maintained at theorganization-wide level.

To reduce costs and to ensure easy scalability, theKM group implemented KShop on five PCs, whichalso acted as servers. Acting on feedback fromemployees, the KM group offered users tools tocustomize their respective KShop entry pages.The KM group also integrated access to corporatedata and several locally managed repositories intoKShop to provide a single entry point to much ofthe codified knowledge within Infosys.

The content on KShop was organized into differentcontent types—for instance, BOKs, case studies,reusable artifacts, and downloadable software—with each content type having its own home page.Every knowledge asset under a content type wasassociated with one or more nodes (representingareas of discourse) in a knowledge hierarchy ortaxonomy. Multiple paths were created throughthe hierarchy to facilitate easy categorization andretrieval of tagged knowledge assets. As the num-ber of knowledge assets and nodes proliferatedwith time, information overload became a distinctpossibility. To address this problem, the KM groupinitiated efforts to fine-tune its categorizationscheme and make it more relevant to the differentpractice communities. A member of the KM groupreflected on these efforts.

For us, taxonomy is not just a frameworkfor categorizing content; it is a strategy tounify multiple constituencies. Going for-ward, the search engine will be enhancedto leverage the taxonomy for deliveringaccurate search results. For this ap-proach, we need a taxonomy that is moreelaborate than the current one. The nextversion of KShop will support automaticclassification tools. Even with the taxon-omy being huge, this means easierclassification for users.

Learnability

“Knowledge is the currency of the new millenniumand we are building a company that will remain atthe forefront of knowledge management.” TheCEO of Infosys offered this assessment in 2000 ashe reflected on the role that knowledge has playedin transforming a little known company into aglobal player within two decades. Operating in thehighly dynamic software services market withclients distributed around the world, Infosyscontinues to place an emphasis on leveraging itsemployees’ knowledge for corporate advantage.As the company’s chairman and chief mentorfrequently observes, “Our key assets walk out ofthe door every evening, and it is themanagement’s responsibility to see that theyreturn the next morning.” Not surprisingly, Infosysis among the few companies in the world thatvalues and reports its human capital on its balancesheet (for specific details, see Raghuram 2001).

How does Infosys build its human capital? As withother companies, Infosys recruits bright peopleand trains them regularly. Yet, given the speedand complexity of change that its employeesconfront, Infosys realized that formal training alonewould not suffice for its employees to remain at thecutting edge of software development and deploy-ment. Not only would the time lag betweentraining and actual application compromise perfor-mance, but exclusive reliance on training alsocould detract from employees’ ability to innovate atthe point of knowledge deployment.

For these reasons, the company began recruitingemployees for their “learnability.” At Infosys, learn-ability refers to an “ability to derive generic con-clusions from specific instances of learning.” Inthis sense, learnability is much more than refiningexisting knowledge through a process of learning-by-doing. The director of human resources atInfosys clarified further:

The only thing that is constant in thisindustry is change. If we want our peopleto address this change, it does not matterwhether they know specific technologieslike C++ or Java. That is something we



can teach. More important is whetherthey are able to figure out how Java issimilar to or different from C++ and makeappropriate adjustments in applying it.Or, having solved a problem for one cus-tomer, can they apply that knowledge ina generic way to some other problem thatthey face later? This is why we recruitpeople who possess this generic learningcapability that we call learnability.

Learnability is manifest in a noticeable tendencyamong Infosys employees across levels andfunctional areas to think and speak in terms ofmodels. These models are bundles of assump-tions, constructs, experiences, and workinghypotheses ranging from the customer relationshipmodel, which defines the way Infosys employeesinteract with customers, to an iterative model ofsoftware development, which encourages con-tinual feedback and adjustments during projectimplementation (Jalote 2000, p. 74). Even thegenesis of the company-wide KM program can betraced to a knowledge maturity model thatevaluates the maturity level of knowledge pro-cesses (for more details, see Kochikar et al. 2002).

By no means are these models static templateswhose only purpose is to transfer knowledge fromone context to another. Rather, they are dynamicentities that coevolve with employees’ experi-ences. Such coevolution is critical for employeesto progress up the career ladder as they adaptfrom one job paradigm to another. At an organi-zational level too, learnability has played a vitalrole in the transition that Infosys employees havemade from a predominantly Y2K-driven businessmodel to one driven by e-commerce. The knowl-edge maturity model, which so far has guided KMefforts at Infosys, is itself being modified toincorporate new lessons gained during the imple-mentation of the company-wide KM program.

Informal Communities andFormal Workgroups

To ensure that knowledge created by employeesbenefits their colleagues, Infosys encourages the

formation of rich social networks among em-ployees. Within these networks, knowledgesharing occurs informally with employees callingcolleagues for help, thereby engendering an“asking culture” at Infosys. More recent manifes-tations of this asking culture include e-mailbroadcasts for help on specific topics and theposting of queries on online bulletin boards ordiscussion groups. As an associate vice presidentwho has climbed up the ranks explained,

Information goes around informally. I cancall up someone to get answers. Or, Ican post a query or send an email and Iwill not be at all surprised to get severalresponses within five or ten minutes fromcolleagues located around the world. Westill have a campus-like environment,though this may change as the companygrows bigger.

To strengthen the firm’s rich informal networks, theKM group developed a tool called the PeopleKnowledge Map (PKM) in 1998. The PKM, de-ployed on the corporate intranet, catalogs thenames and contact information of internal expertsin specific areas, thereby enabling colleagues tolocate them and benefit from their expertise easily.The PKM forges connections between com-munities and their respective knowledge bases.This tool is especially useful to the constant streamof newcomers who are not familiar with thepockets of expertise distributed within the knowl-edge network at Infosys.

Unfolding in parallel are formal processes withinproject teams. These project teams conceive,design, and complete software projects—the coreof the value proposition offered by Infosys. Withineach team, the creation and exchange ofknowledge is governed by the strong bonds forgedamong team members as they work long hourstogether under intense time pressure. A core ofexperienced members always remains with theteam even as other members are rotated to otherproject teams. These senior members mentornewcomers on idiosyncratic technologies, tools,and client requirements.



Each project team is organized into modules, witheach module dealing with one aspect of a complexproject. By 2000, most project teams had em-braced an iterative model of software develop-ment. The iterative model is a fluid, adaptiveprocess for the development of complex softwarein a rapidly changing environment. Rather thanrelying on sequential deployment of resources andactivities, the iterative model employs paralleldeployment. Rather than different modules withineach team working in their own knowledge spaces,the iterative model forces overlap betweenmodules reinforced by continual interaction andfeedback. Such overlap enables members of agiven module to specialize in particular tasks but,at the same time, have some general knowledgeof the tasks performed by team members in othermodules. In their attempts to explain this structureand development process, both a project managerand a senior developer offered the human brain asan analogy to describe how their project teamsfunctioned.

Organizational Routines

To provide a template for routines for knowledgeaccumulation, Infosys adopted the capabilitymaturity model (CMM). Developed by the Soft-ware Engineering Institute at Carnegie-MellonUniversity, the CMM gauges the maturity level ofa software company’s processes and method-ologies on a scale of 1 to 5. Each of the five levelshas built into it a series of steps that allow asoftware company to accumulate the knowledgeand experience to move sequentially from onelevel to the next. As the company advances to thenext level, additional steps force further reflectionand improvement. At Level 5, the level at whichInfosys operates, a company not only hasmechanisms to prevent defects and manage tech-nological change, but also the ability to quantify,measure, and continually modify its softwaredevelopment processes (for more details, seeJalote 2000).

For instance, implementation of CMM at Infosysincludes a mechanism to enable its project teamsto learn from completed projects. Through audits,members of a project team identify what went rightor wrong during the course of a project. Moreimportantly, a closure report written at the end ofeach project captures important lessons for thefuture. Typically, these closure reports includeitems such as the duration of the project,resources employed and other facts that allow afuture reader to gauge the efficiency and effec-tiveness with which the project was implemented.These reports also contain a section on causalanalysis that records major deviations in processperformance and lists possible causes for thesedeviations. At the end of this report, a conclusionsummarizes the major points learned from theproject for future reference. Closure reports serveas a key mechanism linking knowledge creationand deployment at the work group level with therest of the organization.

The flexibility embodied in the CMM frameworkenables Infosys to try new initiatives, learn fromthem, assimilate the outcomes, and, in the pro-cess, change its very processes and routines. Inthis sense, Infosys’ CMM Level 5 induced organi-zational routines are analogous to learnability,which drives knowledge creation and deploymentat the individual level. Over time, Infosys hasadopted the CMM framework not just for softwaredevelopment, but also for all other organizationalinitiatives. For instance, an Infosys regionaldirector offered a specific instance of how thecompany has applied the CMM-inspired iterativeimplementation process to an initiative other thanthe fine-tuning of software methodologies.

When we started the first off-campusDevelopment Center within India, it wasa revolutionary step for us. We startedout on a very small scale….We wentthrough several issues and problems,and we committed mistakes….At theend, after several experiments over a fullyear, we came out with a very scalable



and repeatable process to set up devel-opment centers. We went through thesame piloting process when we startedour first Development Center outsideIndia. Today, we have the capability toset up development centers anywhere inthe world just like that.

Catalyzing the Knowledge Spiral

By the beginning of 2000, Infosys appeared tohave put together the necessary elements of aknowledge system at each organizational level. Ithad recruited employees for learnability and devel-oped informal processes and formal structures toenhance knowledge creation and sharing. It hadleveraged CMM Level 5 routines as the frameworkfor organization-wide learning and change.Furthermore, in implementing the central knowl-edge portal KShop, it had created a digital platformfor the accumulation and reuse of organizationalknowledge.

These initiatives were not sufficient by themselvesto jump-start a virtuous knowledge circle. Patron-age of KShop by employees remained low.Employees within various project teams andpractice communities continued to use theirinformal networks to access knowledge in times ofneed. Local repositories of specialized knowledgecontinued to proliferate within project teams andpractice communities. In other words, processesat different levels of the knowledge system werenot coupling and reinforcing one another.

In response, during the first quarter of 2001, theKM group implemented a major initiative—theknowledge currency unit (KCU) incentivescheme—to jumpstart contributions to KShop.Under the scheme, Infosys employees whocontributed or reviewed contributions to KShopwould be awarded KCUs, which they couldaccumulate and exchange for monetary rewards orprizes. Additionally, employees’ cumulative KCUscores would be displayed on a scoreboard onKShop, thereby increasing the visibility andstanding of prolific contributors.

Intended and UnintendedConsequences

These initiatives began yielding results, especiallyafter the KCU incentive scheme was introduced.For instance, within a year of introduction of theKCU scheme, over 2,400 new knowledge assets—project proposals, case studies, and reusablesoftware code—were contributed to KShop, withnearly 20 percent of Infosys employees contri-buting at least one knowledge asset. Over130,000 KCUs were generated by the KM groupand distributed among contributing and reviewingemployees.

Even as these events unfolded, the KM groupbegan wondering if the KCU incentive scheme hadbecome too successful. One concern had to dowith employees experiencing information overloadand, consequently, higher search costs forreusable knowledge. As a member of the KMteam commented,

If the repository becomes too heavy, thechances of getting useful information re-duce with time. So, there is a trade-offthat people have to make, especiallybecause we are looking at increasinglyshort life-cycle projects—nowadays, 6weeks to 3 months. Suppose someonesearches the repository, gets three docu-ments, takes 2 or 3 days to read thesedocuments and finds out that they are notuseful. Then, he might question the verypoint of searching the repository, con-sidering it a waste of time….Somepeople have told us informally that theyare finding it faster to do things on theirown or to ask someone they knowinstead of searching the repository forreusable content.

Complicating matters, the explosive growth in thenumber of contributions began placing a heavyburden on the limited number of volunteerreviewers. A shortage of reviewers made it difficultfor the KM group to ensure that contributions werereviewed for quality and relevance before beingpublished on KShop. With review processes still



struggling to keep pace with the accelerating paceof contributions, assets of uncertain quality beganappearing on KShop. When even contributions ofquestionable quality began receiving high qualityratings from colleagues, the rating scheme itselfcame under scrutiny. A manager commented,

Our experience is showing that relyingsolely on incentives may not be the rightway to increase knowledge sharing.Incentives increase awareness and thenumber of contributions. But, the qualityof these contributions is in questionbecause some people are gaming thesystem….Then, there are groups withinthe company that have a sharing cultureand don’t care about incentives. Thenumber of contributions generated bythese groups is as much or more thanthe rest of the company put together.

Concerns also began emerging about the possibleimpacts of the KCU scheme on knowledgeprocesses at the other levels of the organization.One such concern was the potential for the KCUincentive scheme to destroy the spirit of com-munity and the asking culture within the company.What employees would have given freely to eachother earlier was now being monetized through theKCU incentive scheme. “Why not gain somerewards and recognition for my knowledge con-tributions, especially when others are doing so?”was the question being asked by employees whohad shared their knowledge earlier for free for the“joy of sharing.”

An additional concern was the real possibility thatsome project teams and practice groups,disappointed with KShop, could revert to buildingand relying on their own local repositories insteadof contributing to the central portal. A projectmanager explained that this trend could result inthe fragmentation of the knowledge commons.

Nowadays, there are many useful knowl-edge assets being retained at the teamor practice unit levels that never make itto KShop. There is a growing impressionthat many units are holding their assets

close to their people in local repositories.With time, this may become a barrier totrue knowledge sharing or reuse.

Taken together, these concerns and unanticipatedemergent processes had the potential to compro-mise the key objective of the company-wide KMprogram: to make every instance of learningwithin Infosys available to every employee. Amanager who had been associated with the KMinitiative from the beginning reflected on thesechallenges.

We are coming to realize that knowledgemanagement requires much more thanjust technology. We have to pay atten-tion to the cultural and social facets ofknowledge management as well. Wehave to continually campaign and evan-gelize besides investing the time andresources to manage the content.Knowledge management initially appearsto be a deceptively simple task. But,make just one wrong move and it isdifficult to convince people to come back.

Process Interventions

Sensing the potential of the KCU incentive schemeto compromise the company-wide KM program,the KM group took prompt action. First, theyintervened to decouple knowledge sharing fromthe economic incentives that threatened the spiritof community and the perceived utility of KShop.Specifically, in April 2002, the KM group modifiedthe KCU incentive scheme to emphasize recog-nition and personal visibility for knowledge sharingcontributions more than monetary rewards. Itformulated a new composite KCU score thatemphasized the usefulness and benefit of contri-butions to Infosys as rated not just by volunteerreviewers or colleagues, but also by actual users.Moreover, to increase the accountability ofreviewers and users who rated contributions toKShop, the KM group began demanding tangibleproof to justify any high ratings. Finally, the KMgroup significantly reduced the number of KCUs



awarded for reviewing contributions to KShop andraised the bar for cashing in the KCU incentivepoints for monetary rewards. The KM grouphoped that these steps would shift the motivationto share knowledge away from monetary rewards.

A second set of initiatives focused on improvingKM practices within project teams and practicecommunities. Intense time pressure in completingprojects within stringent deadlines reduced knowl-edge codification efforts within teams. To addressthis issue, the KM group modified forms and pro-ject templates to facilitate extraction of knowledgeusing automated tools. The group also imple-mented a project-tracking tool on KShop to logdetails and deliverables pertaining to every projectwithin Infosys. The objective of these initiativeswas to enable the codification and extraction ofknowledge even as teams carried out their routineproject-related tasks.

Despite these attempts, knowledge codificationcontinued to vary across project teams. To ad-dress this shortcoming, the KM group introduceda hierarchy of roles to broker knowledge sharingbetween project teams, practice communities, andthe wider organization. Within each project team,one volunteer member would be designated as theKM prime. The KM prime would be responsible foridentifying and facilitating the fulfillment of theteam’s knowledge needs for each project. The KMprime would also ensure that, after the completionof each project, the team codified and sharedcritical knowledge gained during the project withthe rest of the company. At the practice com-munity and wider organizational levels, the KMgroup also created the role of knowledge cham-pions to spearhead and facilitate knowledgesharing and reuse in critical or emerging tech-nologies and methodologies. Furthermore, the KMgroup encouraged employees to swap stories onKShop with the view of promoting widespreadsharing of tacit individual and team-level knowl-edge and experiences.

These initiatives certainly had an impact. After themodified KCU scheme was introduced, those whohad contributed to KShop just to secure monetaryrewards reduced their participation. For instance,

in the two quarters immediately following theintroduction of the modified KCU scheme, thenumber of new contributors per quarter declinedby nearly 37 percent, whereas the number of newknowledge assets contributed to KShop per quar-ter declined by approximately 26 percent duringthe same period. After this significant initialdecline, however, the number of new knowledgeassets contributed to KShop slowly stabilized andthen increased at a more manageable pace.Users of KShop reported lower search costs andsignificant increases in the quality and utility ofknowledge assets available through the portal.Looking into the future, there was also muchoptimism that the KM prime and knowledgechampion roles would yield positive outcomes.

These initiatives underscore the continual natureof change at Infosys. The KM program at Infosyscontinues to evolve based on feedback fromInfosys employees and the KM group’s continualefforts to gauge the effectiveness of their variousinitiatives. As Infosys continues to grow in termsof its work force, geographical reach, and valueproposition, new challenges will surely emerge.Reflecting on the transformative nature of changethat shapes the company and its KM program, acompany director pointed out,

Many years ago, people would ask, “Areyou sure where you are going? Do youknow what issues you will get into?” Ouranswer to these questions is still thesame: “No, but we have the processesin place to address these issues as andwhen they arise. And, as we addressthese issues, we will transform our-selves.”

Virtuous Circles, Vicious Circlesand Steering

A systems way of thinking (Maruyama 1963;Masuch 1985; Perrow 1984; Senge 1990; Weick1969) provides us with a theoretical perspective tounderstand these dynamics. First, it enables a



deeper understanding of how an organizationmight attempt to generate a virtuous knowledgecircle through initiatives at and across levels toachieve a dynamic balance between forces forcontinuity and change. Next, it explains how andwhy the very initiatives taken to generate avirtuous circle may also end up generating avicious circle. Finally, it also offers insights intoprocess interventions that knowledge managersmight use to steer their organization’s knowledgesystem around or out of vicious circles and toenhance the potential for virtuous circles toemerge (see Figure 2 for a summary).

Virtuous Circles

The Infosys case suggests how an organizationcan accomplish a dynamic balance by institu-tionalizing practices at and across different organi-zational levels. At the individual level, recruitingemployees for their ability to reflect-in-actionbalances the tendency to engage only in knowl-edge refinement through learning-by-doing. AtInfosys, employees endowed with learnabilityencapsulate their experiences in models. Suchmodels serve both as models of and also forknowledge experiences (Geertz 1973). In additionto channeling learning efforts, these models, whenapplied to new contexts, enable employees toengage in both single-loop and double-looplearning and generate new knowledge.

At the group level, interlaced structures provide thebenefits of knowledge from communities as well asfrom workgroups. These interlaced structuresforce epistemic overlap between members of dif-ferent communities—what Nonaka and Takeuchi(1995) label as shared division of labor. Rich con-nections between different workgroup modules,each subscribing to different epistemologies, allowknowledge from across these communities torecombine, thereby generating innovative solutionsto emergent problems. At the same time, sucharrangements afford mutual control which pro-duces an ongoing mediated consensus (Polanyi1966, p. 72).

Key to the balancing act at the collective level isthe recognition that volume can overwhelm valuewithin digital repositories (Brown and Duguid 2002,p. xiii). In this regard, Infosys developed categori-zation schemes to enable easy search and re-trieval of knowledge assets from digital reposi-tories. As these categorization schemes are cus-tomizable by different user communities to betterreflect their respective thought-worlds, the poten-tial for inconsistencies between induced and emer-gent categories is minimized. At a process level,these initiatives are reflective of adaptive struc-turation, wherein rules inscribed in technologiesand rules constitutive of social processes coevolve(DeSanctis and Poole 1994; Giddens 1986;Orlikowski 1992).

Organization-wide routines—in the case of Infosys,the capability maturity model—forge a dynamicbalance between the sustenance of core compe-tencies and the onset of core rigidities. CMMoffers a template to pilot initiatives, learn fromexperience, and iteratively scale up only thoseinitiatives that prove successful. Accordingly,Infosys’ implementation of CMM illustrates howorganizations might leverage routines as sourcesof both continuity and change (Feldman andPentland 2003) to develop dynamic capabilitiesover time (Eisenhardt and Martin 2000; Teece etal. 1997).

These institutionalized practices are all necessarybut by themselves not yet sufficient to generate avirtuous knowledge circle. An additional require-ment is the coupling of knowledge processesacross different levels to jumpstart the estab-lishment of a knowledge commons. In this regard,models of collective action (Schelling 1978;Gladwell 2000) demonstrate that a critical thresh-old has to be crossed for a bandwagon to emerge.Recognizing that socio-psychological processesmay prevent this critical threshold from beingreached, Oliver et al. (1985) have highlighted theneed for incentives to create a bandwagon.

Infosys instituted several initiatives to coupleknowledge processes unfolding across the dif-ferent levels. For instance, the People Knowledge



Company-wide KM program,KShop portal implemented

KCU incentive scheme introduced to jump start

contributions to KShop portal

• Employees articulate what they know

• Digital options

• KShop portal complements asking culture

• Robust knowledge commons

• Employees articulate more than they know

• Information overload

• Incentivized sharing threatens asking culture

• Potential fragmentation of knowledge commons

Increasing returnsvirtuous circle

Market-for-knowledgevicious circle

KCU incentive scheme modifiedKM Prime role

Project tracking tool

Steering

Company-wide KM program,KShop portal implemented

KCU incentive scheme introduced to jump start

contributions to KShop portal

• Employees articulate what they know

• Digital options

• KShop portal complements asking culture

• Robust knowledge commons

• Employees articulate more than they know

• Information overload

• Incentivized sharing threatens asking culture

• Potential fragmentation of knowledge commons

Increasing returnsvirtuous circle

Market-for-knowledgevicious circle

KCU incentive scheme modifiedKM Prime role

Project tracking tool

Steering

Figure 2. Vicious and Virtuous Circles at Infosys

Map was implemented to couple processes acrossthe individual and group levels. The projectclosure report initiative is illustrative of initiatives tocouple processes across the group and collectivelevels. In addition, the KCU scheme was animportant initiative to couple processes across theindividual and collective levels of the organization.

The catalyzing effect of incentives in generating abandwagon was all too evident at Infosys, withcontributions to KShop increasing significantlyafter the introduction of the KCU scheme. Indeed,it seemed as though Infosys had successfullyinitiated a virtuous knowledge circle. Why then didpotentially negative consequences arise forInfosys’ knowledge system? To address thisquestion, it is useful to look at the darker side ofmutually causal processes. The very samemutually causal processes that have the potentialto generate a virtuous circle can just as easilygenerate a vicious one.

Vicious Circles

Vicious circles are triggered when feedback gener-ated at a particular system level is amplified acrossthe entire system, setting in motion events thatgenerate unintended negative consequences(Maruyama 1963; Masuch 1985; Senge 1990;Weick 1969). Especially in systems with tightlycoupled components (Orton and Weick 1990), asin the case of an organization’s knowledge system,mutually causal feedback loops can easily beamplified across the system, thereby rendering itmore susceptible to pathologies (Perrow 1984).

Market-for-knowledge vicious circle. This wasthe case with Infosys’ intervention to jumpstartcontributions to KShop through the KCU incentivescheme. The incentives worked in that contri-butions to KShop increased significantly. Unfor-tunately, however, contributions began increasingfaster than the system’s ability to review them for



quality. Moreover, some employees were soincentivized that they began articulating “morethan they knew.” The resulting information over-load, together with the decreasing quality of knowl-edge assets available on KShop, increased searchcosts for users and affected reuse adversely.

This sequence of events bears out observationsmade by Hansen and Haas (2001) that attention—not information—is a scarce resource. Employeeswere incentivized to articulate their knowledge, andarticulate they did. Ensuing dynamics led toinformation overload on KShop threatening todisrupt the very virtuous circle that Infosys hadgenerated with considerable effort.

Senge (1990) conceptualized such situations asexhibiting dynamic complexity. Dynamic com-plexity is inherent in

situations where cause and effect aresubtle, and where the effects over time ofinterventions are not obvious….When thesame action has dramatically differenteffects in the short run and the long,there is dynamic complexity. When anaction has one set of consequenceslocally and a very different set of conse-quences in another part of the system,there is dynamic complexity. Whenobvious interventions produce non-obvious consequences, there is dynamiccomplexity (Senge 1990, p. 71).

These observations were certainly true of thepattern of relationships at and across the differentorganizational levels within Infosys. Local actionsat each organizational level had global conse-quences. Short-term results were different fromlong-term results. Indeed, in real time, interven-tions such as the KCU scheme appeared to be theobvious ways to proceed, but the non-obviousoutcomes, such as information overload, couldonly be understood over time.

Other potential vicious circles. It is not difficultto think of other vicious circles an organizationmay confront as it attempts to keep its knowledge

system in dynamic balance. For instance, con-sider the connections between tacit and explicitknowledge. Excessive emphasis on explicatingand codifying knowledge can create severalpathologies. We have already alluded to theinformation overload that may emerge when tacitknowledge is first explicated and then stored indigital repositories. In addition, the very articu-lation of tacit knowledge can end up trivializing it(Polanyi 1966, p. 20). As Tsoukas (1996, p. 14)noted, “individual knowledge is possible preciselybecause of the social practices within whichindividuals engage—the two are mutually defined.”As a result, efforts to codify knowledge in anabstract form to enable wider reuse may make itmore difficult for colleagues to apply such knowl-edge across contexts. In other words, codificationmay paradoxically reduce knowledge reuseinstead of increasing it.

Consider another vicious circle. Organizationswould surely like to recruit employees for theirability to be reflective practitioners. However, suchemployees may prefer to create knowledge anewas they deal with problems, instead of reusing theknowledge created by others and stored in digitalrepositories. In other words, hiring bright individ-uals who can generate new knowledge mightparadoxically reduce knowledge reuse from digitalrepositories and the potential for increasing returnsaccruing from such reuse.

Although these vicious circles are only illustrative,they highlight certain properties of the knowledgesystem. First, the effects of initiatives taken at onelevel of the system can be felt across differentlevels. Second, these effects feed back into thesystem and may get amplified due to the mutuallycausal nature of processes unfolding at and acrosslevels. Third, effects of specific initiatives are notimmediately obvious because of time lags betweencauses and consequences. As a result, theresolution of a particular problem at a given levelor time may create a different problem at anotherlevel or time.

These observations highlight a key paradox ofknowledge management: that an organization’sknowledge system contains seeds of its own



destruction. Leave it alone, and virtuous knowl-edge circles may never materialize. Intervene tocouple processes at and across different levels,and vicious circles are bound to emerge.

Steering the Knowledge System

Given these dynamics, what role should knowl-edge managers play in supporting their organi-zation’s knowledge system? An answer to thisquestion requires an appreciation of the mutuallycausal processes that constitute an organization’sknowledge system. Despite the almost axiomaticnature of this statement, many knowledgemanagers continue to think in terms of straightlines when “reality is made up of circles” (Senge1990, p. 70). Consequently, many of their inter-ventions are based on a linear view of rela-tionships between variables wherein changes inone element of a system are expected to lead to aproportionate change in another (Mohr 1982).According to Weick (1969, p. 81), “managerscontinue to believe that there are such things asunilateral causation, independent and dependentvariables, origins, and terminations.”

In situations characterized by dynamic complexity,as is the case with an organization’s knowledgesystem, solutions based on a linear way of thinkingcan often exacerbate the problem instead ofsolving it. Specifically, a change in one part of asystem can have a disproportionate impact on adifferent part of the system in a subsequent timeperiod and the interactions between the parts cangenerate negative outcomes. By the time suchoutcomes are understood, the system has oftenalready locked itself into a vicious circle.

To handle mutually causal processes, therefore,organizational interventions need to be processual(Massey et al. 2002, p. 287). In other words,interventions need to address process drivers andthe ways in which these drivers interact with oneanother over time (Drazin and Sandelands 1992;Pettigrew 1992; Tsoukas 1989). As Senge (1990)pointed out, this mindfulness entails seeingbeyond local detail complexity to identify dynamiccomplexity in the broader knowledge system.

We offer steering4 as a processual way forknowledge managers to address these dynamics.Just as experienced drivers switch from cruisecontrol to active steering at busy intersections orcongested roadways, knowledge managers needto proactively anticipate emerging pathologieswithin the knowledge system and steer aroundthem. Steering also implies an ability to extricatean organization that inadvertently finds itself miredin a vicious circle.

Steering around vicious circles. To steer,knowledge managers must first develop sensitivityto the dynamic complexity inherent in theirorganization’s knowledge system and to the onsetof vicious circles (Senge 1990). This requires anepistemology that recognizes the web of mutuallycausal processes constituting the knowledgesystem. It also means forsaking the traditionallinear view of understanding phenomena in termsof necessary and sufficient causation (Mohr 1982).

Such a shift in mindset redirects attention to theinherently distributed and diverse nature of knowl-edge processes across different levels of anorganization (Hutchins 1995). An implication isthat knowledge management cannot be centra-lized in one person. No one person can possessthe diversity of perspectives and the cognitiveability to interface with the many distributed andmutually causal knowledge processes constitutingthe knowledge system. Instead, consistent withthe principle of requisite variety (Ashby 1965;Morgan 1986; Shannon and Weaver 1949),management of knowledge within an organizationis best left distributed among a team of individualswith diverse epistemic leanings.

Steering out of vicious circles. An organizationmay find itself trapped in a vicious knowledgecircle despite steering. As Masuch (1985, pp. 22-23) noted,

4In using the term steering, we have been influenced bythe work of Kemp et al. (2001) on strategic nichemanagement and policymaking as a process of socio-technical change.



Vicious circles lead an absurd existencesince everyone should avoid “deviation-amplifying”5 feedback. Yet, once caughtin a vicious circle, human actors continueon a path of action that leads further andfurther away from the desired state ofaffairs.

Likewise, with regard to such vicious circles,Kanter (1977, p. 249) observed that “it is hard fora person to break out of the cycle once begun.”

How, then, might knowledge managers steer theknowledge system out of a vicious circle? Oneway would be for them to identify and decouplesystem processes that may have triggered thevicious circle (Starbuck 1996; Weick 1969).Actions by Infosys to decouple the associationbetween monetary incentives—its KCU incentivescheme—and its employees’ knowledge behaviorsis an illustration of decoupling. Such decouplingbreaks the deviation amplifying feedback loopsdriving the vicious circle, thereby affording knowl-edge managers an opportunity to steer out of it.

Knowledge managers could also introducedeviation counteracting feedback loops throughinterventions in other parts of the system.Deviation counter-acting feedback loops arrest thetendency of the system to drift further and furtheraway from the desired outcome (Masuch 1985;Senge 1990). In the case of Infosys, the institutionof the KM prime and knowledge champion rolesand automated tools for extracting knowledge fromredesigned forms and project templates constituteefforts to counteract the negative impact of timepressure on the extent of knowledge codification.

In summary, a systems view of knowledgemanagement sensitizes knowledge managers tothe fact that vicious circles may emerge despiteand even because of their best efforts. At thesame time, it affords them the potential to dyna-

mically steer around or out of vicious circles whenthey arise. In doing so, it offers an epistemologythat departs from approaches that either grantknowledge managers primacy over organization-wide processes or afford them no such agency.

Implications and Conclusion

Knowledge is key to the continued vitality oforganizations, but managing knowledge as anorganization-wide resource is not easy. What is itabout knowledge that entices yet entraps thosewho try to manage it for increasing returns? Ourin-depth analysis of events and experiences atInfosys offers several insights into the nature of thechallenges that organizations confront in har-nessing knowledge. First, an organization’s knowl-edge system comprises mutually causal processesthat unfold at and across different organizationallevels. Second, these mutually causal processesgenerate opposing forces that need to be balanceddynamically to generate a virtuous circle. Third, anorganization’s knowledge system contains seedsof its own destruction, as the very initiatives thatthe organization undertakes to generate a virtuouscircle have the potential to generate vicious circlesas well. Fourth, knowledge managers must inter-vene processually to steer their organization’sknowledge system around and out of viciouscircles that are bound to emerge.

Underlying these insights is a systems view oforganizational knowledge. Such a systems viewopens up new avenues of research on knowledgemanagement. For instance, consider studies thatexplore specific approaches to building an organi-zation’s knowledge system. Among others, theseinclude (1) an approach to knowledge creation thatstresses the role of individuals, (2) a communitiesof practice approach that emphasizes informalrelationships based on shared language andthought-worlds, and (3) a repositories-based ap-proach that emphasizes codification and centralstorage of organizational knowledge. From asystems perspective these different approachesare constituent pieces of an organization’s knowl-

5Deviation amplifying feedback progressively leads asystem further and further away from intended outcomes.Therefore, it increases the deviation between intendedoutcomes and realized outcomes over time.



edge system rather than stand-alone pieces.From such a perspective, it would be instructive toexplore how these constituent pieces interact withone another to enable or impede the generation ofvirtuous circles. More specifically, it would beinteresting to explore the differential conditions thatcreate complementarities or substitutive effectsamong knowledge derived from repositories, com-munities, and creative individuals.

Indeed, a systems perspective offers a wealth ofopportunities to explore and mitigate specific ten-sions that may arise within and across organi-zational levels. For instance, consider the impactof knowledge codification on reuse. Explicationthrough codification has the potential to divorce thecodified knowledge from its context, therebyinhibiting the propensity of employees to reuseknowledge from organizational digital repositories.How might knowledge be represented to enhancethe propensity of employees to reuse codifiedknowledge from digital repositories? Or, considerthe effect of time and work pressures on knowl-edge management processes and outcomes.Such pressures may reduce employees’ propen-sity to share information with one another. In sucha case, how may technological tools, work prac-tices, and social mechanisms be integrated toalleviate the tensions that time and work pressuresgenerate?

At its core, a systems perspective offers a certainepistemology for conducting research on knowl-edge management. First, by focusing our attentionon mutually causal processes and dynamiccomplexity, a systems perspective shifts theemphasis of research to an exploration of pro-cesses and their drivers. In doing so, it under-scores the importance of employing longitudinalapproaches to research. Second, a systemsperspective raises the possibility that, despitemanagement’s best efforts, vicious circles are justas likely to emerge as virtuous circles.Accordingly, it sensitizes researchers to thepossibility of unanticipated negative outcomes inthe context of knowledge management. Only if wepay attention to these facets can we fullyappreciate the challenges and potential ofmanaging knowledge as an organization-wideresource.

Acknowledgements

We are indebted to Nandan Nilekani and the manyemployees at Infosys Technologies who gener-ously gave of their time and insights. Our specialthanks to the members of Infosys KnowledgeManagement group, especially Dr. J. K. Suresh,C. S. Mahind, Mahesh Venugopalan, and AnoopKarunakaran. We thank Daniel Buenza, RogerDunbar, Sanjay Jain, Natalia Levina, V. Samba-murthy, and Kim Wade-Benzoni for their com-ments on earlier versions of this paper. We alsothank Neha Bajaj, Shelley Rescober, and MonicaMalhotra for research assistance and Neha Bagchiand Joshua Krantz for editorial help. Thereviewers and associate editor at MIS Quarterlyoffered critical and generous feedback that hasshaped this paper. We are indebted to them.

References

Alavi, M., and Leidner, D. “Review: KnowledgeManagement and Knowledge ManagementFoundations and Research Issues,” MISQuarterly (25: 1) 2001, pp. 107-136.

Argote, L. Organizational Learning: Creating,Retaining, and Transferring Knowledge, Klu-wer Academic Publishers, Boston, MA, 1999.

Argyris, C., and Schon, D. OrganizationalLearning: A Theory Of Action Perspective,Addison Wesley, Reading, MA, 1978.

Arrow, K. “The Economic Implications of Learningby Doing,” Review of Economic Studies(29:3), 1962, pp. 155-173.

Ashby, W. R. An Introduction to Cybernetics,Methuen and Co Ltd., London, UK, 1965.

Berger, P. L., and Luckman, T. The Social Con-struction of Reality: A Treatise in the Socio-logy of Knowledge, Doubleday, New York,1967.

Bowker, G. S., and Star, S. L. Sorting Things Out:Classification and its Consequences, MITPress, Cambridge, MA, 2000.

Brown, J. S., and Duguid P. “OrganizationalLearning and Communities of Practice:Toward a Unified View of Working, Learning,and Innovation,” Organization Science (2:1),1991, pp. 40-57.



Brown, J. S., and Duguid, P. The Social Life ofInformation, Harvard Business School Press,Boston, 2002.

Callon, M. “The Embeddedness of EconomicMarkets in Economics,” in The Laws of theMarkets, M. Callon (Ed.), Blackwell Pub-lishers, Oxford, UK, 1998, pp. 1-57.

Columbia Accident Investigation Board. ReportVolume 1, National Aeronautics and SpaceAdministration, Washington, D.C., 2003.

Davenport, T. H., and Prusak, L. Working Knowl-edge: How Organizations Manage WhatThey Know, Harvard Business School Press,Boston, 1998.

David, P. A. “Clio and the Economics ofQWERTY,” Economic History (75), 1985, pp.227-332.

DeSanctis, G., and Poole, M. S. “Capturing theComplexity in Advanced Technology Use:Adaptive Structuration Theory,” OrganizationScience (5:2), 1994, pp. 121-147.

Dougherty, D. “Interpretive Barriers to SuccessfulProduct Innovations in Large Firms,” Organi-zation Science (3:2), 1992, pp. 179-202.

Drazin, R., and Sandelands, L. “Autogenesis: APerspective on the Process of Organizing,”Organization Science (3:2), 1992, pp.230–249.

Dutton, J. M., and Thomas, A. “Relating Tech-nological Change and Learning by Doing,” inResearch on Technological Innovation, Man-agement, and Policy Volume 2, R. D. Rosenbloom (Ed.), JAI Press, Greenwich, CT, 1985,pp. 187-224.

Eisenhardt, K. M., and Martin, J. M. “DynamicCapabilities: What Are They?,” StrategicManagement Journal (21), 2000, pp. 1105-1121.

Fahey, L., and Prusak, L. “The Eleven DeadliestSins of Knowledge Management,” CaliforniaManagement Review (40:3), 1998, pp. 265-276.

Faraj, S., and Sproull, L. “Coordinating Expertisein Software Development Teams,” Manage-ment Science (46), 2000, pp. 1554-1568.

Feldman, M., and Pentland, B. “Reconcep-tualizing Organizational Routines as a Sourceof Flexibility and Change,” AdministrativeScience Quarterly (48), 2003, pp. 94-118.

Fortune. "Now Capital Means Brains Not Bucks"January 1991, pp. 31-32.

Fukuyama, F. Trust: The Social Virtues and theCreation of Prosperity, The Free Press, NewYork, 1995.

Garud, R. “Commentary: The Process of Rela-tional Contracting,” in Advances in StrategicManagement Volume 10, J. Dutton, A. Huff,and P. Shrivastava (Eds.), JAI Press,Greenwich, CT, 1994, pp. 137-144.

Garud, R., and Kotha, S. “Using the Brain as aMetaphor to Model Flexible Productive Units,”Academy of Management Review (19), 1994,pp. 671-698.

Garud, R., Kumaraswamy, A., and Malhotra, M.“Infosys: Architecture of a Scalable Corpora-tion,” Case Study, Stern Business School,New York University, 2003.

Garud, R., and Nayyar, P. “TransformativeCapacity: Continual Structuring by Inter-Tem-poral Technology Transfer,” Strategic Man-agement Journal (15: 5), 1994, pp. 365-385.

Geertz, C. The Interpretation of Cultures, BasicBooks, New York, 1973.

Giddens, A. The Constitution of Society: Outlineof the Theory of Structuration, University ofCalifornia Press, Berkeley, CA, 1986.

Gladwell, M. The Tipping Point: How Little ThingsCan Make a Big Difference, Little, Brown,Boston, MA, 2000.

Glaser, B., and Strauss, A. The Discovery ofGrounded Theory: Strategies of QualitativeResearch, Wiedenfeld and Nicholson,London, UK, 1967.

Gold, A., Malhotra, A., and Segars, A. “Knowl-edge Management: An Organizational Capa-bilities Perspective,” Journal of ManagementInformation Systems (18:1), 2001, pp. 185-214.

Govindarajan, V., and Gupta, A. The Quest forGlobal Dominance: Transforming Global Pre-sence Into Global Competitive Advantage,Jossey-Bass, San Francisco, 2001.

Grant, R. M. “Prospering in Dynamically-Com-petitive Environments: Organizational Capa-bility as Knowledge Integration,” OrganizationScience (7), 1996a, pp. 375-387.

Grant, R. M. “Toward a Knowledge-Based Theoryof the Firm,” Strategic Management Journal(17), Spring 1996b, pp. 109-122.



Grover, V., and Davenport, T. “General Per-spectives on Knowledge Management:Fostering a Research Agenda,” Journal ofManagement Information Systems (18:1),2001, pp. 5-21.

Hansen, M. T., and Haas, M. R. “Competing forAttention in Knowledge Markets: ElectronicDocument Dissemination in a ManagementConsulting Company,” Administrative ScienceQuarterly (46:1), 2001, pp. 1-28.

Hargadon, A. How Breakthroughs Happen: TheSurprising Truth About How CompaniesInnovate, Harvard Business School Press,Boston, 2003.

Hayek, F. A. “The Use of Knowledge in Society,”American Economic Review (35:4), 1945, pp.519-532.

Hutchins, E. Cognition in the Wild, MIT Press,Cambridge, MA, 1995.

Jalote, P. CMM in Practice: Processes forExecuting Software Projects at Infosys,Addison-Wesley Publishing Co., Boston,2000.

Jelinek, M., and Schoonhoven, C. B. The Inno-vation Marathon: Lessons from High Tech-nology Firms, Basil Blackwell Ltd., Oxford,UK, 1990.

Kanter, R. M. Men and Women of the Corporat-ion, Basic Books, New York, 1977.

Kemp, R., Rip, A., and Schot, J. “ConstructingTransition Paths Through the Management OfNiches,” in Path Dependence and Creation,R. Garud and P. Karnøe (Eds.), LawrenceEarlbaum Associates, Mahwah, NJ, 2001, pp.269-302.

Kochikar, V. P. “Knowledge—The Currency of theNew Millennium,” Infosys White Paper, 2001.

Kochikar, V. P., Mahesh, K., and Mahind, C. S.“Knowledge Management in Action: TheExperience of Infosys Technologies”, inKnowledge and Business ProcessManagement, V. Hlupic (Ed.), Idea GroupPublishing, Hershey, PA, 2002, pp. 83-98.

Kogut, B., and Zander, U. “Knowledge of theFirm: Combinative Capabilities and the Repli-cation of Technology,” Organization Science(3:3), 1992, pp. 383-397.

Latour, B. “Technology is Society Made Durable,”in A Sociology of Monsters. Essays on

Power, Technology and Domination, J. Law(Ed.), Routledge, London, 1991, pp. 103-131.

Lave, J., and Wenger, E. Situated Learning: Le-gitimate Peripheral Participation, CambridgeUniversity Press, Cambridge, UK, 1994.

Leidner, D. “Editorial,” Journal of StrategicInformation Systems (9), 2000, pp. 101-105.

Leonard, D. Wellsprings of Knowledge, HarvardBusiness School Press, Boston, 1998.

Leonard-Barton, D. “Core Capabilities and CoreRigidities: A Paradox in Managing New Pro-duct Development,” Strategic ManagementJournal (13), Summer 1992, pp. 111-125.

Levina, N. "Collaborative Practices In InformationSystems Development: A Collective Reflec-tion-in-Action Framework," in Proceedings ofthe 23rd International Conference on Infor-mation Systems, L. Applegate, R. D. Galliers,and J. I. DeGross (Eds.), Barcelona, Spain,2002, pp. 267-277.

Levitt, B., and March, J. G. “OrganizationalLearning,” Annual Review of Sociology (14),1988, pp. 319-40.

Lincoln, Y. S., and Guba, E. G. NaturalisticInquiry, Sage Publications, New York, 1985.

March, J. G. “Exploration and Exploitation inOrganizational Learning,” OrganizationScience (2:1), 1991, pp. 71-87.

March, J. G., and Simon, H. A. Organizations,John Wiley, New York, 1993.

Markus, L. “Toward a Theory of KnowledgeReuse: Types of Knowledge Reuse Situa-tions and Factors in Reuse Success,” Journalof Management Information Systems (18:1),2001, pp. 57-93.

Markus, L., Majchrzak, A., and Gasser, L. “ADesign Theory for Systems that SupportEmergent Knowledge Processes,” MIS Quar-terly (26:3), 2002, pp. 179-212.

Maruyama, M. “The Second Cybernetics: Devia-tion Amplifying Mutual Causal Process,”American Scientist (51:2), 1963, pp.164-179.

Massey, A., Montoya-Weiss, M., and O’Driscoll, T.“Knowledge Management in Pursuit of Perfor-mance: Insights from Nortel Networks” MISQuarterly, (26:3), 2002, pp. 269-289.

Masuch, M. “Vicious Circles In Organizations,”Administrative Science Quarterly (30:1), 1985,pp. 14-33.



Miles, M., and Huberman, M. A. Qualitative DataAnalysis: A Source Book of New Methods,Sage Publications, Beverly Hills, CA, 1984.

Miller, K. “Knowledge Inventories and ManagerialMyopia,” Strategic Management Journal (23),2002, pp. 689-706.

Mohr, L. B. Explaining Organizational Behavior:The Limits and Possibilities of Theory andResearch, Jossey-Bass, San Francisco,1982.

Morgan, G. Images of Organization, Sage Publi-cations, Beverly Hills, CA, 1986.

Nelson, R., and Winter, S. An EvolutionaryTheory of Economic Change, HarvardUniversity Press, Cambridge, MA, 1982.

Nidumolu, S., Subramani, M., and Aldrich, A.“Situated Learning and the Situated Knowl-edge Web: Exploring the Ground BeneathKnowledge Management,” Journal of Man-agement Information Systems (18:1), 2001,pp. 115-150.

Nonaka, I., and Takeuchi, H. The Knowledge-Creating Company, Oxford University Press,New York, 1995.

Okhuyzen, G. A., and Eisenhardt, K. M. “Inte-grating Knowledge in Groups,” OrganizationScience (13), 2002, pp. 370-386.

Oliver, P., Marvell, G., and Teixeira, R. “A Theoryof the Critical Mass: Interdependence, GroupHeterogeneity and the Production ofCollective Action,” American Journal ofSociology (91), 1985, pp. 522-566.

Olsen, M. The Logic of Collective Action: PublicGoods and the Theory of Groups, HarvardUniversity Press, Cambridge, MA, 1965.

Orlikowski, W. J. “The Duality of Technology:Rethinking the Concept of Technology inOrganizations,” Organization Science (3:3),1992, pp. 398-427.

Orlikowski, W. J. “Knowing In Practice: Enactinga Collective Capability in Distributed Organi-zing,” Organization Science (13: 3), 2002, pp.249-273.

Orr, J. E. “Sharing Knowledge, CelebratingIdentity: Community Memory in a ServiceCulture,” in Collective Remembering, D.Middleton and D. Edwards (Eds.), SagePublications, London, 1990, pp. 168-189.

Orton J. D., and Weick K. E. “Loosely CoupledSystems: A Reconceptualization,” Academyof Management Review, (15:2), 1990, pp.203-223.

Penrose, E. The Theory of the Growth of theFirm, Oxford University Press, Oxford, UK,1995.

Perrow, C. Normal Accidents, Princeton Univer-sity Press, Princeton, NJ. 1984.

Pettigrew, A. M. “Character and Significance ofStrategic Process Research,” Strategic Man-agement Journal (13), 1992, pp. 5-16.

Polanyi, M. The Tacit Dimension, Anchor Books,Garden City, NY, 1966.

Raghuram, S. “Management of Human Assets atInfosys,” Case Study, Fordham University,2001.

Raghuram, S., Garud, R., Wiesenfeld, B., andGupta, V. “Factors Contributing to VirtualWork Adjustment,” Journal of Management(27), 2001, pp. 383-405.

Sambamurthy, V., Bharadwaj, A., and Grover, V.“Shaping Agility Through Digital Options:Reconceptualizing the Role of InformationTechnology In Contemporary Firms,” MISQuarterly (27:2), 2003, pp. 237-263.

Sandelands, L. E., and Stablein R. E. “TheConcept of Organization Mind,” in Researchin the Sociology of Organizations Volume 5,S. Bacharac and N. DiTomaso (Eds.), JAIPress, Greenwich, CT, 1987, pp. 135-161.

Schelling, T. C. Micromotives and Macrobehavior,W. W. Norton and Company, New York, 1978.

Schultze, U., and Leidner, D. “Studying Knowl-edge Management In Information SystemsResearch: Discourses and Theoretical As-sumptions,” MIS Quarterly (26:3), 2002, pp.213-242.

Schon, D. A. The Reflective Practitioner, BasicBooks, New York, 1983.

Senge, P. M. The Fifth Discipline, Doubleday/Currency, New York, 1990.

Shannon, C., and Weaver, W. The MathematicalTheory of Communication, University ofIllinois Press, Champaign, IL, 1949.

Shapiro, C., and Varian, H. Information Rules,Harvard Business School Press, Boston,1999.



Soo, C., Devinney, T., Midgley, D., and Deering,A. “Knowledge Management: Philosophy,Processes and Pitfalls,” California Manage-ment Review (44:4), 2002, pp. 129-150.

Spender, J. C. “Making Knowledge the Basis of ADynamic Theory of the Firm,” Strategic Man-agement Journal (17), Winter 1996, pp. 45-62.

Starbuck, W. H. “Unlearning Ineffective orObsolete Technologies,” International Journalof Technology Management (11:7/8), 1996,pp. 725-737.

Subramaniam, M. and Youndt, M. A. “The Influ-ence of Intellectual Capital on the Types ofInnovative Capabilities,” Academy of Manage-ment Journal, 2004 (forthcoming).

Swidler, A. “Culture In Action: Symbols andStrategies,” American Sociological Review(51:2), 1986, pp. 273-286.

Teece, D. A., Pisano, G., and Shuen, A. “DynamicCapabilities and Strategic Management,”Strategic Management Journal (18), 1997, pp.509-534.

Tsoukas, H. “The Firm as a Distributed Knowl-edge System: A Constructionist Approach,”Strategic Management Journal (17), Winter1996, pp. 11-25.

Tsoukas, H. “The Validity of Idiographic ResearchExplanations,” Academy of ManagementReview (14), 1989, pp. 551–561.

Von Krogh, G., Ichijon, K., and Nonaka, I.Enabling Knowledge Creation: How to Unlockthe Mystery of Tacit Knowledge and Releasethe Power of Innovation, Oxford UniversityPress, New York, 2000.

Walsh, J. P., and Ungson, G. R. “OrganizationalMemory: Information Acquisition, Retention,and Retrieval,” Academy of ManagementReview (16), 1991, pp. 57-91.

Weick, K. E. The Social Psychology ofOrganizing, Addison-Wesley, Reading, MA,1969.

Weick, K. E., and Roberts, K. “Collective Mind InOrganizations: Heedful Interrelating on FlightDecks,” Administrative Science Quarterly(38:3), 1993, pp. 357-381.

Winter, S. G. “Knowledge and Competence asStrategic Assets,” in The Competitive Chal-lenge: Strategies for Industrial Innovation andRenewal, D. J. Teece (ed.), Ballinger, Cam-bridge, MA, 1987, pp. 159-184.

Zuboff, S. In the Age of the Smart Machine, BasicBooks, New York, 1984.

About the Authors

Raghu Garud is on the faculty at the Stern Schoolof Business, New York University. He is coeditorof Organization Studies and an associate editor ofManagement Science. Currently, Raghu is co-editing (with Cynthia Hardy and Steve Maguire) aspecial issue on Institutional Entrepreneurship toappear in Organization Studies.

Arun Kumaraswamy is a visiting associateprofessor of Management at the Lee Kong ChianSchool of Business, Singapore Management Uni-versity, Singapore. Arun works in the areas ofstrategy and technology/innovation managementand has published in journals such as the StrategicManagement Journal and the Academy of Man-agement Journal. Along with Raghu Garud andRichard Langlois, he is coeditor of Managing in theModular Age published by Blackwell in 2003.

Garud & Kumaraswamy/Vicious and Virtuous Circlesphp.scripts.psu.edu/users/r/u/rug14/12.Viscious and...

Documents

Transcript of Garud & Kumaraswamy/Vicious and Virtuous Circlesphp.scripts.psu.edu/users/r/u/rug14/12.Viscious and...