technology daptation the case of a omputer-supported inter ...majchrza/virtualteamsinmisq.pdf ·...

Majchrzak et al./Technology Adaptation

MIS Quarterly Vol. 24 No. 4, pp. 569-600/December 2000 569

RESEARCH ARTICLE

TECHNOLOGY ADAPTATION: THE CASE OF ACOMPUTER-SUPPORTED INTER-ORGANIZATIONALVIRTUAL TEAM1

By: Ann MajchrzakInformation and Operations ManagementMarshall School of BusinessUniversity of Southern CaliforniaLos Angeles, CA [email protected]

Ronald E. RiceSchool of CommunicationInformation and Library StudiesRutgers UniversityNew Brunswick, NJ [email protected]

Arvind MalhotraKenan-Flagler Business SchoolUniversity of North Carolina, Chapel HillChapel Hill, NC [email protected]

Nelson KingSchool of EngineeringUniversity of Southern CaliforniaLos Angeles, CA [email protected]

1Sirkka Jarvenpaa was the accepting senior editor forthis paper.

Sulin BaInformation and Operations ManagementMarshall School of BusinessUniversity of Southern CaliforniaLos Angeles, CA [email protected]

Abstract

The adaptation process for new technology is notyet well understood. This study analyzes how aninter-organizational virtual team, tasked withcreating a highly innovative product over a 10month period, adapted the use of a collaborativetechnology and successfully achieved its chal-lenging objectives. The study of such a virtualteam is especially useful for extending our under-standing of the adaptation process as virtualteams have more malleable structures than typicalorganizational units and controlled group experi-ments. Data were obtained from observations ofweekly virtual meetings, electronic log files, inter-views, and weekly questionnaires administered toteam members. We found that the team initiallyexperienced significant misalignments among thepre-existing organizational environment, group,and technology structures. To resolve thesemisalignments, the team modified the organiza-tional environment and group structures, leavingthe technology structure intact. However, as theteam proceeded, a series of events unfolded that


570 MIS Quarterly Vol. 24 No. 4/December 2000

caused the team to reevaluate and further modifyits structures. This final set of modifications in-volved reverting back to the pre-existing organi-zational environment, while new technology andgroup structures emerged as different from boththe pre-existing and the initial ones. A new modelof the adaptation process—one that integratesthese findings and those of several previousmodels—is proposed.

Keywords: CBCS, collaborative work systems,groups, problem solving, information attributes,longitudinal study, remote work, innovation

ISRL Categories: HA08, HA12, AA09, AC03,AD05, AI0113, BD103, DD05

Introduction

This study analyzes how an inter-organizationalvirtual team, tasked with creating a highly inno-vative product over a 10 month period, adaptedthe use of a collaborative technology and suc-cessfully achieved its challenging objectives. Wewere interested in understanding what adaptationsoccurred: Were they primarily with the techno-logy, work group, or organizational environment?How often did these adaptations occur? Whatwas the role of pre-existing structures in theseadaptations? The study of a virtual team isespecially useful for extending our understandingof the adaptation process as virtual teams havemore malleable structures than typical organiza-tional units and controlled group experiments.The paper proceeds by first extending structura-tion theory using the context of a virtual team,then describing our case study research method-ology and results, and ending by discussing thetheoretical and practical implications of theresults.

Conceptual Development

In this section, we first summarize structurationtheory and several models for explaining theadaptation process, identifying key differencesbetween the models. Resolving these differences

forms the basis for our research questions. Weargue that virtual teams provide a “revelatory”case, where a researcher has an opportunity tostudy a previously inaccessible phenomenon, orat least, as of this time, a unique case (Yin 1994)in which to address our questions. We elaboratethis point by introducing the virtual team we hadthe opportunity to study.

Structuration Theory Applied toInformation Systems

Structuration theory, largely associated withGiddens’ (1984) institutional theory of socialevolution, has been used to explain organizationaladoption of computing and information tech-nologies (Barley 1986; Orlikowski 1992; Orlikowskiand Robey 1991; Rice 1994; Rice and Gattiker1999). Adaptive structuration theory extendsstructuration models to consider mutual influenceof technology and social processes (DeSanctisand Poole 1994; Poole and DeSanctis 1990).

Structuration theory suggests that the imple-mentation and use of new technology are notdeterministic; technologies are structured by usersin their contexts of use (Contractor and Eisenberg1990; DeSanctis and Poole 1994; Johnson andRice 1987; Orlikowski 1992; Orlikowski and Yates1994; Poole and DeSanctis 1990; Rice andGattiker 1999; Walsham 1993; Yates andOrlikowski 1992). The structuring of technologiesin use refers to the processes through which usersmanipulate and reshape their technologies toaccomplish work and the ways in which suchaction draw on the particular social contexts withinwhich they work.

Based on research conducted from a structurationtheory perspective, the technology adaptationprocess is now understood to be one that evolvesover time—sometimes gradually, sometimesdiscontinuously—in response to interruptions(Tyre and Orlikowski 1994) or intentionalmanagement policy (Johnson and Rice 1987;Orlikowski et al. 1995), and is constrained by pre-existing structures (Barley 1986) of the organi-zation and its associated tasks, technology, andthe group (DeSanctis and Poole 1994). Thus, in


MIS Quarterly Vol. 24 No. 4/December 2000 571

Technology’s StructuralFeatures and Spirit

Task and OrganizationalEnvironment

Group’s Internal Structure

AppropriationMoves andFaithfulness

DecisionProcesses

DecisionOutcomes

New SocialStructures

Emergent Sources ofStructures

Technology’s StructuralFeatures and Spirit

Task and OrganizationalEnvironment

Group’s Internal Structure

AppropriationMoves andFaithfulness

DecisionProcesses

DecisionOutcomes

New SocialStructures

Emergent Sources ofStructures

Figure 1. Summary of Adaptive Structuration Model (DeSanctis and Poole 1994)

the context of these pre-existing structures, newtechnology represents “occasions for restruc-turing,” not determinants of particular outcomes.

There are several different models in the literaturefor describing how the adaptation process unfolds.Adaptive Structuration Theory, as described byDeSanctis and Poole, portrays the process bywhich technologies are adapted as consisting ofstructures, appropriations, and decision out-comes. As Figure 1 summarizes, their modeldescribes three sources of structures as pre-existing conditions that form the context in whichthe technology is implemented and, as such,affect appropriations, which in turn affect decisionprocesses and outcomes. Technology structuresinclude the restrictiveness, sophistication, andcomprehensiveness of its features as well as thetechnology’s “spirit,” the general intent of thetechnology with regard to values and goals. Taskand organizational environment refers to thenature of the task (such as complexity and inter-dependence) and the organizational setting suchas hierarchy, corporate information, and culturalbeliefs. The group’s structure includes the inter-action patterns and decision-making processes ofits members.

Appropriations, which may be subtle and difficultto observe, are defined as the immediate, visibleactions that evidence deeper structuration pro-

cesses. Assessment of appropriation processes isat the heart of the Adaptive Structuration Theoryframework, by documenting exactly how techno-logy structures are being invoked for, orconstrained in use in, a specific context. Appro-priations can be analyzed for their faithfulness (theextent to which appropriations are in line with thetechnology’s spirit), their instrumental uses, or theusers’ attitudes. One hypothesis proposed byDeSanctis and Poole is that the more faithful theappropriation—i.e., the more that appropriationsalign with the technology’s initial intent—the morelikely the team’s decision processes will lead tosuccessful outcomes.

Figure 2 summarizes Leonard-Barton’s (1988)model of the (successful) adaptation process.Here, the technology adaptation process is seenas cycles of misalignments, followed by align-ments, followed by more but smaller misalign-ments, gradually evolving to a state in which thetechnology, the delivery system and the perfor-mance criteria are aligned. In this model, suchfactors as initial unfaithfulness to a technology’sspirit are less detrimental to outcomes thanleaving misalignments unresolved. This role ofmisalignments is similar to Tyre and Orlikowski’snotion of discrepant events in the technologyadaptation process. They found that discrepantevents might be helpful for a group as they maytrigger technology adaptations that in the long run



Alignment

User environment

Technology

Misalignments in-technology-delivery system-performance criteria

cycles

AlignmentAlignment

User environmentUser environment

TechnologyTechnology



cycles

Figure 2. Summary of Misalignment Model (Leonard-Barton 1988)

make the technology more useful to the organi-zation. However, the slope of adaptation,according to Tyre and Orlikowski, is not gradual,but initially steep with only brief windows ofopportunity in which technologies could bemodified. This role of discrepant events intriggering adaptations is similar to the process ofreinvention proposed by Johnson and Rice,whereby the innovation choices are seen to beinitially matched to (and often constrained by)prior adoption agendas, but may then be adaptedin terms of use or form.

These various models of adaptation, therefore, allagree that adaptation is a process of modifyingexisting conditions in an effort to achievealignment. The models disagree, however, on thenature of this adaptation. First, what is changedduring the adaptation process to bring aboutalignment, according to both the Leonard-Bartonand the Tyre and Orlikowski models, are any andall structures. In contrast, structuration theory(Giddens 1984) explicitly states that existingstructures represent (varying) constraints on theadaptation process and thus are not necessarilyall adapted equally. DeSanctis and Poole suggestthat at least one structure—technology spirit—does not change during the adaptation processand poses a constraint on adaptation. Barley

suggests a different structure—political and statusdifferences—that poses a constraint on adapta-tion. We suggest that these differences betweenthe models may be artifacts associated with thefield research and controlled experimental settingsthat have had to examine technologies or politicalstructures not amenable to change (such asinformation systems imposed on users). Thus, atheoretical conclusion that technology spirit orpolitical and status differences are unlikely tochange may be more a reflection of these limitson research designs than of theoretical con-clusions about what can occur.

Empirically, it may be that a structure—anystructure—can constrain an adaptation processnot by virtue of what it is (e.g., technology, politics,status), but by the simple reality of it not beingmalleable, and that malleability may be socontext-specific that a priori recommendationsidentifying particular structures as more or lessconstraining have the effect of inappropriatelyfocusing attention on that particular structure.Thus, to resolve this difference between themodels about whether any or only some struc-tures constrain adaptation, one must examine asituation in which the structures are as malleableas possible to see which ones indeed change andwhich ones force other structures to adapt.



The models differ in another way as well: in theimportance attributed to the degree or amount ofmisalignments. When DeSanctis and Poole saythat “faithful” spirits lead to successful group out-comes, they imply that the fewer and smaller theadaptations, especially with regard to spirit, themore successful the outcome. In contrast, Tyreand Orlikowski propose that as long asmisalignments can eventually be accommodated,the number or type of misalignments do notmatter. The concept that “fewer is better” may bea reflection of a field study reality that often onlypermits few and small adaptations, or is focusedon a large, transformative change, rather than atheoretical conclusion about what is likely to leadto success. Thus, examining a situation in whichthe structures are malleable both in the amountand degree of adaptation would permit addressingthis difference between the models.

Finally, the models differ in the continuitypresumed to occur in the adaptation process.Leonard-Barton’s model proposes that adapta-tions occur continuously in response to misalign-ments, gradually leading to a successful align-ment. In contrast, Tyre and Orlikowski charac-terize adaptation as a highly discontinuousprocess, where discontinuities occur during briefwindows of opportunity which open the constraintset. This difference may reflect different condi-tions in the field rather than invariant theoreticalconclusions. For example, adaptations may notbe discontinuous by nature but reflect the fieldsetting that Tyre and Orlikowski studied, i.e., onein which structures only became malleable atdiscontinuous intervals. Putting those samestructures in a different field setting may haveyielded a continuous adaptation process ratherthan a discontinuous one. Thus, the adaptationprocess may be neither inherently discontinuousnor continuous but rather responsive to changesin structural malleability, whenever that mayoccur.

Research Questions

The above differences in models suggest theneed for further research on the adaptationprocess. The ideal research site would be one inwhich all structures would be as malleable as

possible, yet occur in a real-world context so thatexternal validity is maintained. In such a context,a workgroup that is permitted to modify itsstructures at the outset of a task allows us toaddress the difference in the models about whichstructures are malleable; a workgroup that ispermitted to make both small and significantadaptations allows us to address the differencesin the models about the amount and degree ofadaptation; and a workgroup that is permitted tomake changes throughout the process of taskaccomplishment allows us to address thedifferences in the models about continuous anddiscontinuous change. In such a setting, then,four research questions become salient:

(1) Can the workgroup adapt any or allstructures, or does it primarily try to adapt tothe technology’s initial spirit?

(2) Do pre-existing structures constrain theworkgroup’s adaptation process, even whenthese structures are malleable?

(3) After the initial adaptation to achievealignment, does the workgroup experiencethe need for further adaptations?

(4) What is the nature of these adaptations: arethey discontinuous, responding to windows ofopportunities, or are they continuous, gra-dually closing misalignments?

A Context for Minimizing Constraintsof Existing Structures: A Virtual Teamand Collaborative Technology

We had the opportunity to observe a newlyconstituted inter-organizational virtual teamresponsible for developing a revolutionary newproduct while adapting to a new collaborativetechnology. Virtual teams are defined as “groupsof geographically and/or organizationally dis-persed coworkers that are assembled using acombination of telecommunications and informa-tion technologies to accomplish an organizationaltask,” which may be temporary and thus adaptiveto organizational and environmental changes(Townsend et al. 1998, p.18). Because a virtual



team spans multiple organizational contexts, thevirtual team and sponsoring organizations areoften more willing to experiment with new struc-tures and thus offer an ideal opportunity toobserve the results of malleability in theirstructures (Bowers 1995; Grudin 1994; Jarvenpaaand Leidner 1998; Townsend et al. 1998).

The team we studied, named SLICE (Simple,Low-cost, Innovative Concepts Engine), consistedof eight engineers from three different organi-zations: five from RocketCo, two from SigmaCo,and one from StressCo. The project was jointlyfunded by the three companies. Team membersdevoted 10 months, at about 15% of their time, tothe project. The team members were selected torepresent expert knowledge in a specific disciplinerelated to the task; they had not worked togetherpreviously. Virtual team members were geo-graphically distributed: two members were locatedin different ends of the same building, three othermembers were each one mile away in differentbuildings; one member of a second organizationwas located 100 miles away; and two members ofthe third organization were located 1,000 milesaway in different buildings. SLICE memberslimited their travel since they were involved inmany different teams within their company. As aresult, all members were together only once—atthe end—although there were three other formalmeetings held in which some members attendedand some informal conversations among mem-bers of one organization. The combination ofgeographic dispersion of team members, imposeddisciplinary heterogeneity, different organizationalaffiliations, and lack of historical working rela-tionships required the team to create structures forovercoming the challenge of sharing knowledge orsimply understanding each other (Grudin 1994;Jarvenpaa and Ives 1994; Jarvenpaa and Leidner1998; McKenney et al. 1992; Mohrman et al.1995; Purser et al. 1992).

In addition to these common characteristicsshared by virtual teams, the team we studiedfaced additional challenges not commonlyencountered by a virtual team: it was tasked todevelop (but not build) a concept and drawings fora revolutionary and highly complex rocket designthat could be marketed by the three companies.RocketCo executive management, in charge of

overseeing the technical quality of the work,recognized the revolutionary nature of the designtask upon seeing the initial specifications for thenew rocket design and voiced extreme skepticismthat such a design could be created. Sinceproduct components were tightly coupled, teammembers from all three companies needed towork in highly interdependent iterative virtualbrainstorming sessions, a structure with prece-dence for neither the team members nor the com-panies. In addition, inter-organizational collabora-tion for initial concept development is alsounprecedented since new product development isconsidered a competitive advantage for each ofthese companies independently; other organi-zations are traditionally only included indevelopment efforts after an initial concept isdevised. The application of virtual teams to con-ceptualizing a new product added another layer ofchallenge and newness to the process. Recog-nizing the unique challenges faced by the team,management gave the team wide latitude tochange structures. Thus the SLICE team providedan excellent opportunity to study a team with moreopportunities for malleability than normal.

Virtual teams are made possible in large part dueto rapid developments and diffusion of colla-borative technologies (CTs). CTs include, at aminimum, a virtual workplace that provides arepository recording the process of the group,electronic information-sharing (such as throughfile sharing, e-mail, and electronic whiteboards),meta-information on the entries in the repository(such as date, sequence, and author of eachcontribution), and easy access and retrieval fromthe repository (Romano et al. 1998). As a result,such systems facilitate the access, creation,processing, storage, retrieval, distribution, andanalysis of information across positional, physical,and temporal boundaries and allow the incor-poration of members from other units andorganizations with specific, otherwise difficult-to-obtain expertise (Lipnack and Stamps 1997;Warkentin et al. 1997).

The CT used for the SLICE project was called the“Internet Notebook” (www.nexprise.com). Whilethe team members had experience withrudimentary CTs (e-mail, file transfer, videoconferencing), the Notebook was seen as the first



Table 1. Features of the Internet Notebook

Characteristics of CTs Features of the Internet Notebook

Interface A custom-designed HTML browser that allowed members of the virtualteam distributed access over a private network.

Different forms of inter-action (such as e-mail andelectronic white-boards)

The CT allowed the team members to simultaneously work on an entrytogether as a distributed team: to author new documents (called entries),comment on entries, and draw sketches as entries. This system waseventually complemented by using telephone conferencing along withsynchronous system entries for synchronous, multi-media collaboration.

Shared information stor-age, access, and retrievalthrough meta-information(such as date, sequence,and author of each contri-bution to the repository)

The CT allowed team members to remotely access the Notebook fromanywhere; to sort entries (such as by date, keywords, or reference links);to retrieve entries (such as by navigating or filtering entries by keyword,subject, date, or author as well as seeing the network of reference links tofind frequently referenced entries for a particular topic). Team memberscould create a personal profile for e-mail notification of relevant entries andhad a document vault in which documents requiring configuration controlwere entered.

A record of the process ofthe group

The CT allowed members creating new entries to make explicit referencelinks in an entry to prior entries; it had entries that contained summary ofteleconference meetings, identifying action items and decisions made.

Libraries of solutions andpractices

Team members could create and use templates for re-occurring teamactivities (such as minutes, agendas, and action items).

CT suitable for a complex engineering designcollaboration because it supported multiple mediatypes across multiple platforms for enteringengineering content (i.e., graphical entries). Thevirtual team was asked to pilot the Notebook aspart of their work effort. Table 1 lists specificfeatures of the Notebook corresponding to thegeneral characteristics of CTs noted above.

Research Methodology andData Collection

Researchers have suggested that observations bemade of the micro-processes of adaptation overtime to determine which types of adaptation aremore likely to lead to successful group outcomes(Barley 1986; Orlikowski and Gash 1994). Inaddition, such studies should not be limited toshort time spans of technology use, as adapta-

tions may occur over time. Finally, such studiesshould avoid obtaining data retrospectively, as thisencourages respondents to gloss over detailsassociated with variations in adaptation eventsover time or they may be biased by the samerecently institutionalized practices and perceptionsthey are attempting to describe. Our data collec-tion methods satisfy the above requirements andwere geared toward understanding under whatcircumstances technological adaptations occur(Griffith 1999; Orlikowski and Gash 1994).

Case Study Methodology

Since emphasis was on understanding the pro-cess of the virtual team adapting the technologyover its 10 month life span, a descriptive casestudy was used (Myers 1997; Walsham 1993).Case study is a well-accepted approach to studythe complex phenomena of technology implemen-



tation in an organizational setting (Alavi andCarlson 1992; Orlikowski and Baroudi 1991; Yin1994). In our data collection effort, we usedinterviews and documentary materials as theprimary source of data (Myers 1997). We con-ducted private interviews with each of the eightmembers of the team at seven points in time: atthe outset of the project to understand the existingand initial structures affecting the team; at 7, 14,21, 28, and 35 week intervals to identify changesoccurring at these times; and at a final meetingwhen a “lessons-learned” session was conducted.We also interviewed the seven RocketCo execu-tive technical managers responsible for projectoversight immediately after a mid-project reviewand again at the end of the project. We inter-viewed the Internet Notebook developer at severaljunctures during the team’s work to determine hisoriginal intentions concerning the use of the tech-nology and his reactions to how the technologywas actually being used. In addition to the inter-views, we also examined the entries created bythe team in the Notebook to see how the CT wasbeing appropriated.

We complemented the above methods with ethno-graphic data collection (Geertz 1973; Harvey andMyers 1995) in order to understand how theteam’s social and cultural structures evolvedduring the adaptation process (Lewis 1985). To dothis, two of the study authors became observers inthe team’s process, listening in on all 89 audio-conferences, reading all 1,000+ entries to theNotebook, and noting all uses of the Notebookduring the teleconferences.

In addition to the teleconference notes, the firstauthor attended all of the four in-person meetingsthat were held during the project. Not all of theteam members attended all of the meetings (thekickoff meeting was attended by five out of theeight members, a mid-project technical designreview by two members, a brainstorming sessionby four members, and a final technical reviewattended by all eight members). At each meeting,copious notes were taken of all conversations,including who said what, as well as additionalsocial and non-verbal cues that were used (e.g.,reference to a Notebook entry, hand gestures,eye-to-eye contact, etc.).

Several researchers have recommended triangu-lating qualitative methods with quantitativemethods to ensure that the richness afforded byqualitative methods is supported by quantitativeanalysis (Gable 1994; Lee 1991; Markus 1994;Williams et al. 1988). Thus, we combined the richnote-taking of the observations, teleconferences,and interviews with a quantitative analysis of theCT entries as well as having team memberscomplete a short weekly questionnaire. The ques-tionnaire asked the percent of time on the projectthat members collaborated with each other thatweek (versus working alone on analysis) and theproportion of the collaboration time in which CTversus interpersonal (face-to-face or phone)media were used.

Finally, at the end of the project, the sevenexecutives who attended the final technical reviewand were responsible for judging the outcome ofthe team’s work, as well as the eight teammembers, were asked to complete a short ques-tionnaire indicating the degree (on a 7-point scale)to which the team had effectively accomplishedeach of the team’s work objectives.

Case Analysis Procedure

The analysis followed four steps. The first stepinvolved reviewing the initial interviews to identifythe pre-existing structures for each team member,i.e., how the team members would typically designa new product for their companies. We refer tothese pre-existing structures as “Structures at T1."Structures were grouped according to theDeSanctis and Poole (1994) classification oftechnology, group, and organization environment.Since the essential nature of the task—newproduct development—did not change throughoutthe project, the task structure was not examined.According to DeSanctis and Poole, technologyincludes both the features and the spirit; spirit isdefined as “the ‘official line’ which the technologypresents to people regarding how to act whenusing the system, how to interpret its features, andhow to fill in gaps in procedure which are notexplicitly specified” (p. 126). Group structuresexamined included roles and actions of teammembers. Organizational environment is definedby DeSanctis and Poole as corporate information,



cultural beliefs, and modes of conduct thatinfluence how a team behaves. For the SLICEteam, executive managers provided this environ-ment by conveying and maintaining organizationalculture, protecting the organization from risk, andpreserving control in the face of an inherentlyunpredictable creative process.

The second step in the analysis was to identifyappropriations and when they occurred. Appro-priations made within the first few weeks of theproject were referred to as “AppropriatedStructures at T2"; those that occurred midway orlater in the project were referred to as “Appro-priated Structures at T3." Appropriations, equi-valent to adaptations, are defined by DeSanctisand Poole as “immediate, visible actions thatevidence deeper structuration processes” (p. 128),or more precisely, changes to structures. Appro-priations were identified by reviewing the notes ofteleconferences, in-person meetings, interviews,and the lessons-learned focus group meeting tofind changes in structures. Changes in structurewere then grouped together into similar appro-priation topics. Eight specific topics were iden-tified and grouped into four general topics:

1. Access to the communication tool (who getsaccess, when should they get access);

2. What knowledge is captured (what knowledgegets captured, how is knowledge captured);

3. What helps knowledge sharing (what isshared, what helps sharing); and

4. How are decisions made (who participates inwhat decisions, are technical requirementsquestioned).

For example, notes on a team member’s dis-cussions about how to simplify navigating thegrowing number of entries in the InternetNotebook were labeled as the appropriation topic,“What helps knowledge sharing?” and groupedwith discussion on which keywords to use whencreating an entry. This labeling of notes as anappropriation topic was then corroborated by areview of the CT entries referenced during the

discussion. For instance, when the team mem-bers discussed the need for accuracy in theirentries during one teleconference (since outsiderswould be reviewing the entries), the researcherschecked the entries the team members hadmentioned to determine how accuracy wasmanifested. In this example, we found that theseentries had no dissension or comments that mightquestion the validity of the results posted in theentry. Any interpretations made about CT entrieswere further corroborated by contacting the authorof the CT entry to determine his intent. In sum,the appropriation topics were not theoreticallydeduced, but explicitly inductively derived. Thesetopics, therefore, are not confirmation of aresearcher’s a priori expectations of which appro-priations would be made, but rather represent theappropriations that team members actually made.

In the third step of the analysis, we examined ournotes to identify reasons why each change hadoccurred. For all reasons we were able to identifyfrom our notes, we contacted individuals at thesites to discuss our reasons and to iterate until weobtained concurrence. The reasons were thenreviewed to identify similar patterns.

In the fourth and final step of the analysis, wereviewed the outcome measures to assess thesuccess of the adaptation process. We referredto the final manager survey, which rated theproject on several outcome criteria, and obtaineda summary statement of the project outcomesfrom a senior member of the executive reviewteam. Table 2 summarizes the methodologiesused for each analysis step.

Case Results

The results are organized by the analysis proce-dure: (1) pre-existing structures (Structures atT1); (2) misalignments and appropriated struc-tures after the first few weeks of the project(Appropriated Structures at T2); (3) appropriatedstructures occurring after the midpoint of the pro-ject (Appropriated Structures at T3); (4) reasonsfor changes in structures; and (5) assessment ofproject outcomes.

Table 2. Timing, Nature, and Objectives of Methodologies Used

Time Methodology Used Use of Data in Analysis

Start of the project Private interviews with the eight members of the team, pro-gram manager, and CT developer

Identify structures at T1

Four in-person meetings, whichsome team members attended(kickoff, mid-project technicaldesign review, brainstorming, andfinal technical review)

The first author attended the four in-person meetings andobserved group processes, taking copious notes

Kickoff meeting used to identifystructures at T1

Two mid-project meetings used toidentify appropriated structures at T2

Final meeting used to identifyappropriated structures at T3

At approximately seven-week inter-vals during the project

Two of the authors interviewed each of the eight team mem-bers and program manager

Identify reasons for appropriations

89 teleconferences during the 10month life span of the project

Two of the study authors were observers in the team’s pro-cess, attending all teleconferences (which were held withmembers in their distributed locations and lasted one houreach), listening in on conversations, observing members'use of the CT, and taking copious notes of the conversationsand uses of the CT

Identify appropriated structures at T2and T3 and reasons for appropriations

Weekly (43 weeks) during the 10month life span of the project

The authors analyzed the actual entries into the CT made byteam members

A short weekly questionnaire was administered to teammembers

Identify appropriated structures at T2and T3 and reasons for appropriations

End of the project The first author facilitated a "lessons learned" focus groupmeeting with all team members plus private interviews witheach team member and program manager.

A questionnaire was administered to team members and theirmanagers.

Determine outcomes of team



Result 1: What Was the Nature ofStructures at T1?

The following narrative organizes the pre-existingstructures by the DeSanctis and Poole (1994)classification of organizational environment, groupstructure, and technology. Space considerationspermits us only to highlight the most critical detailsfor understanding these structures.

Organizational Environment at T1RocketCo is known as a leader in rocket design.The executive managers rose to the top of theorganization because of their longevity with theorganization and their technical ability to developnew products that not only out-performed thecompetition but avoided costly and life-threateningfailures. The RocketCo project leader who fundedthis venture was able to convince RocketCoexecutive management to support this effort byproviding documentation of the need for newproducts that met criteria that, in the past, hadbeen thought unachievable, as well as docu-mentation indicating that other organizationsprovided specialized expertise that RocketCocould benefit from during concept development.An agreement to become involved in the project,however, did not change executive management’sbasic risk-aversive nature. A requirement of nolaunch failures had established a proven but tacitprocess imposed by executive management ofrisk reduction through formal reviews and exten-sive analysis tied to test data. In addition,executive managers characterized themselves as“old style” engineers, preferring to “look anengineer in the eye before I’ll believe him when hesays his rocket won’t blow up on the launch pad.”Thus, in terms of our appropriation topics, theexisting organizational environment was charac-terized by executive managers who made little useof electronic communication tools, preferring todiscuss issues in face-to-face encounters; whosaw their role as to identify and adhere to tech-nical requirements for the design to avoid risk;and, as a result, who held to a very hierarchicalnotion of knowledge-sharing (“only those whoneed to know information should gain access to itand when they get information, I’ll give it to themdirectly”) and decision making (“we work with theproject leader, not with the team”).

Group Structure at T1The typical group structure for new productdevelopment at all three organizations was tohave a lead engineer and several specialists. Thelead engineer provides specialists with para-meters to analyze (e.g., to assess whether or nota material would hold up during initial take-off),and then privately integrate the individual analysisresults provided by each specialist. As a result, ateam’s lead engineer made all of the decisions forthe team and acted as a communication hub. Thespecialists rarely interacted with other specialistsinvolved in the project, allowing the lead engineerto work out the conflicts in private. Consequently,there was no need for specialists to use commontools or even a common spatial geometry to sharedesign models. The “big picture” need only beknown to the lead engineer. As a result, the spe-cialists rarely had the context to suggest beneficialchanges to the overall product and, instead,focused their efforts on producing results thatjustified their positions in a rigorous formal reviewprocess to executive management. The spe-cialists organized their results as graphs, with anyassumptions, interpretations, and conclusionsexplained verbally. Changes to designs resultingfrom these analyses would then be made by thelead engineer on new drawing sketches. The leadengineer maintained a personal project binder, inwhich he would place any material he considereduseful for the project. This typically meant that thebinder contained calculations, finalized sketches,and key analysis results that might be needed toeither justify the design or help identify causes ifproblems arose later. Few conversations weredocumented in these binders; typically, actionitems from occasional meetings of the entireproject team were the only public information thatwas captured.

In sum, this existing group structure can becharacterized as one in which specialists usedtheir own tools; the lead engineer did most of thedocumentation but only in private notebooks andmostly just drawings and minutes; the leadengineer would share knowledge with others on aneed-to-know basis; and team members func-tioned with a hierarchical management structure,submitting ideas and analyses to a lead engineer,who then reported these results to executivemanagement.



Technology at T1To coordinate their work, teams were familiar witha variety of technologies, ranging from e-mail andshared files (FTP servers) to videoconferencingand NetMeeting. Managers from all three organi-zations had experienced significant problems witheach of these technologies and looked forward tohaving the opportunity to use the Internet Note-book. The intention of the Internet Notebook’sdeveloper (a former design engineer at Lock-heed’s Skunkworks), and an intention that wassupported and encouraged by team members atthe three organizations, was nothing short of whatwe will call “ubiquitous computing.” The conceptwas that engineers would now be able to post allanalyses, drawings, conversations, schedules,meeting minutes, action items, technical require-ments, assumptions, design rationale, decisions,and presentations in a common repository foreasy access and retrieval; that this repositorywould help the team not only keep track of its ownprocess but be able to review its progress overtime; and that the team would be able to do all thiswithout the time often wasted in finding people,having face-to-face meetings, or waiting until thenext day to get work done. By making the Note-book repository accessible to all those on theteam from anywhere in the world, the team couldeliminate travel to meetings as well as waiting todiscuss issues by phone or in-person. Moreover,if all the information about the project was in theNotebook, there was little need to wait for formalmeetings to apprise executive management of theteam’s progress; instead, executive managementcould be given access to the repository andencouraged to check out the team’s progress ona frequent basis and provide comments at will.

In sum, this spirit of the technology could be sum-marized as: everyone (team members and mana-gers alike) will use the system asynchronously,everyone will participate in capturing their ownknowledge, all knowledge will be captured andshared continuously, and the powerful searchfeatures of the tool will enable everyone to do theirown searches instead of waiting on others to findinformation. It was not the intent of the developer,managers, or team members that the decision-making structure be changed by the tool; as aresult, there were no decision making aides suchas voting or anonymity.

Result 2: What Was the Nature ofInitial Misalignments with NewTechnology and Appropriationsat T2?

Figure 3 presents a summary of the pre-existingstructures at T1, with organizational environmentin the outer circular, group structure in the nextcircular, and the structure for the new technologyin the next most inner circular. The figure alsoindicates whether the organizational environmentand group structures at T1 were initially mis-aligned with the new technology (those with anasterisk indicate misalignment). Finally, the figureshows the appropriations made at T2 (at thebeginning of project) that were required to resolvethese misalignments by organizing themaccording to the eight appropriate topics identifiedin the analysis and labeling them in brackets.

Apparent from the asterisks in Figure 3, theInternet Notebook’s spirit created severalmisalignments with the pre-existing organizationalenvironment and group structure. Examining theappropriations identified in brackets in Figure 3and elaborated in the more detailed discussionbelow is the finding that, for all misalignments, theteam and executive management initiallyaccepted appropriations that deferred to thetechnology’s spirit. That is, they viewed the spiritof ubiquitous computing to be so compelling thatthey were willing to make all necessaryadjustments in their typical mode of operation tomake it happen.

These appropriations were not imposed on them;instead, each team member and each managermade the decision personally and individually,agreeing that the futuristic spirit offered by thetechnology portrayed a work environment muchmore productive than the current one in whichthey worked. In what follows, these appropriationsare briefly described.

Appropriated OrganizationalEnvironment at T2The most obvious misalignment between theorganizational environment and technology at T1was the expectation that management would usethe tool to communicate with the team, an expec-



III.

KN

OW

LED

GE

SHAR

ING

II. KNOWLEDGE CAPTURE

I. ACC

E S S TO S A M

E CO

MM

U NIC

AT ION

T OO

L

Everyone captures,shares & searches all

knowledge continuously,asynchronously withhierarchical decisionmaking and imposed

constraints

Everyoneuse sametool

* No oneaccessesa common

tool[use same

tool]

*Use no tool(prefer F2F)

*[use tool]

No decision making aids

Hierarchical & centralized

Hierarchical between team and management

EXISTING GROUP

EXISTING ORGANIZATION ENVT

* formal reviewminutes only

[all]

*Minimal (drawings &minutes)

[all]

*By others[by self-intermittently]

*By team leader, onlyafter major events

[by self-continuously]*Need-to-know basis[everything] *Need-to-

knowbasis

[everything]

Allcontinuously

*Use others[own]

*Use othersto find[own]

Do own searches and

add links

IV.DECISION MAKING

a) W

hat h

elps

shar

ing?

b) W

hat g

ets

shar

ed?

a) What gets captured? b) How is it captured?

a) Who?

b) When?

Everyone -continuously

All

No aids

Technical design reqsimplicitly known & unchallenged

Technical design reqsimposed by mgta) Participation

in decision makingb) Questioning

technical reqs

NEW TECHNOLOGY

Asynchronous

*Asynch &Synch

[asynch only]

*Synch only[async]

LEGENDLEGEND

* * = misalignment withtechnology at T1

[ ][ ] = Appropriation at T2

Summary of T2Appropriations

III.

KN

OW

LED

GE

SHAR

ING


I. ACC

E S S TO S A M

E CO

MM

U NIC

AT ION

T OO

L

Everyone captures,shares & searches all

knowledge continuously,asynchronously withhierarchical decisionmaking and imposed

constraints

Everyoneuse sametool

* No oneaccessesa common

tool[use same

tool]

*Use no tool(prefer F2F)

*[use tool]

No decision making aids

Hierarchical & centralized

Hierarchical between team and management

EXISTING GROUP

EXISTING ORGANIZATION ENVT

* formal reviewminutes only

[all]

*Minimal (drawings &minutes)

[all]

*By others[by self-intermittently]

*By team leader, onlyafter major events

[by self-continuously]*Need-to-know basis[everything] *Need-to-

knowbasis

[everything]

Allcontinuously

*Use others[own]

*Use othersto find[own]

Do own searches and

add links

IV.DECISION MAKINGIV.DECISION MAKING

a) W

hat h

elps

shar

ing?

b) W

hat g

ets

shar

ed?


a) Who?

b) When?

Everyone -continuously

All

No aids

Technical design reqsimplicitly known & unchallenged

Technical design reqsimposed by mgta) Participation

in decision makingb) Questioning

technical reqs

NEW TECHNOLOGY

Asynchronous

*Asynch &Synch

[asynch only]

*Synch only[async]

LEGENDLEGEND




LEGENDLEGEND




Figure 3. Misalignment of Existing Structures With New Technology at T1 andAppropriations at T2

tation that was misaligned to how they typicallycommunicated with others (using face-to-faceencounters not mediated by a tool), who theytypically communicated with (the team leader, notall team members), when they communicated withthe team (only in synchronous meetings, notasynchronously), what they communicated (formalminutes only, not informal comments), and theirreliance on others to obtain needed information.Executive managers realized the extensiveness ofthese misalignments but agreed that a continuousreview process might make it possible for mana-gers to identify problems in a team’s technicalsolution set early enough to help the team avoidwasting precious time and analysis resources. As

a result, the seven executive managers who wereresponsible for the technical quality of the team’swork agreed to have the tool installed on theirdesktops and to log-in regularly to review andcomment on the team’s progress. However, themanagement decided not to completely foregoformal reviews; instead they decided to institute aformal review of preliminary concepts (i.e.,sketches) halfway through the team’s progress.Neither a continuous review nor a formal review ofpreliminary concepts had been used before atRocketCo.

To help managers with the continuous review, akeyword, “Management_Review” was created.



Team members were instructed to place in the CTunder this keyword those entries that shouldreceive management review, such as those of amore generally “formal” or polished naturecompared to working entries. Managers were freeto roam the Notebook, nevertheless. In addition,managers were provided individual hands-ontraining on the tool and then encouraged to log-inat will.

Appropriated Group Structure at T2The spirit of the Internet technology createdseveral misalignments in the way the team wouldhave normally performed its task. The notion ofubiquitous computing was misaligned with howthe group typically shared knowledge (no oneused a common tool from which to shareknowledge), who shared the knowledge (this wasthe lead engineer’s responsibility), whatknowledge was shared (just drawings andminutes), who received knowledge (only ifneeded), who found the knowledge (others), andwhen knowledge was shared (synchronously).Despite these misalignments, team membersindividually initially agreed to adapt their workprocesses. Interviews indicated that team mem-bers recognized at the outset the degree ofchange that using the Internet Notebook wouldpose but indicated they joined the team explicitlyto do something different.

At the kickoff meeting, group members discussedthe misalignments and generated a protocol forhow they would like to behave, i.e., their norms ofuse. The protocol essentially accepted the spiritof the technology (ubiquitous computing), in largepart because it painted a picture that was seen asenormously beneficial to the team members: itwould save them from meetings, it would allowthem the flexibility of working on the project whenthey had the time, it would ensure that they did notmisplace valuable data, and it would help themunderstand what everyone else was doing on theteam as they floated back and forth between theirmany projects. They agreed to use the InternetNotebook for all their communications, includingputting an end to all face-to-face conversations.This only presented a potential problem for two ofthe members, who tended to see each other in thelunch room, but the other team members felt sostrongly that they needed to be involved in all

conversations that they persuaded the twomembers to agree to holding only publicdiscussions, made accessible via the Notebook.The team members also agreed to try anddocument everything, from conversations todrawings, from using external application viewersto link to private analytic applications (where theycould continue to use their own analytic models),to using a shared project management scheduleto publicly track everyone’s progress. The teammembers received training on the navigation toolsand agreed to create three keywords for everyentry that they put into the repository to facilitatelater retrieval. Finally, they agreed that allmembers would have access to all information,agreeing not to create notebooks only privatelyaccessible.

Summary of Appropriations at T2The inner circle of Figure 3 summarizes theappropriations made at T2, i.e., within the first fewweeks of the project. In addition, the data fromwhich these descriptions of the appropriations aredrawn are presented in Table 3, column 1.

Apparent from this description of the appropriatedstructures at T2 is that the team found itnecessary to make numerous changes to theexisting structures to resolve the misalignments.Of the three structures, group structures andorganizational environment were changed. Theteam chose to leave the initiating technology spiritintact, despite having the possibility of workingclosely with the tool developer to make changes.That is, they accommodated to the technologyspirit, rather than worked to adapt the technologyspirit to accommodate their practices. Finally, theteam did not make changes in the way decisionswere made, since there were no obvious misalign-ments among the existing structures for decisionmaking.

Result 3: What Was the Nature ofAppropriations at T3?

Figure 4 presents the appropriated structures atT3, i.e., at the completion of the project. Thefigure follows the conventions of Figure 3 exceptthat Figure 4 depicts with symbols whether or not



Table 3. Summary of Appropriations and Discrepant Events

Appropriated Structures at T2 Appropriated Structures at T3 Discrepant events

Who Has Access to Same Communication Tool?

Technology Structure

• Managers log on andinterpret work-in-progress as endresult that lacksquality

• Managers can’t makesense of projectalternatives matrix

• Application viewertakes too long tolaunch for each entry

• Team confrontsproblem of sharinganalysis modelsacross platforms

SPIRIT: Managers and team membersuse common toolFEATURES:• Web browser makes tool accessible

to all• Use external application viewers to

share documents— no need forcommon analysis models

SPIRIT: Restricted access to team onlyFEATURES:• Access limited to team• Screen capture feature added, eliminating

need for application viewer

Group Structure

• Team agrees to use tool as their solecommunication medium

• Team will continue to use ownanalysis models to analyze designs

• Team uses tool as one of manycommunication media

• Team uses tool for both communicationand as a common analysis model

Organizational Environment

• Managers agree to use tool to reviewteam’s progress

• Managers agree to NOT use tool

When Do They Access Same Communication Tool?


• Team members toobusy to logon once aday or to entermaterials ahead oftime; only logon inmeeting

• Need synchronousinteraction to makesense of entries

• Unplanned face-to-face encountersoccur

SPIRIT: Leverage for asynchronous useFEATURES:• Server available 24 x 7

SPIRIT: Leverage for synchronous use asmuch as possibleFEATURES• Couple the tool with audio

teleconferences

Group Structure

• Team agrees to logon once a day anddo work asynchronously

• Team agrees to avoid face-to-faceconversations

• All work that requires team decisionmaking is done synchronously

• Face-to-face is OK if results documentedin notebook later on


• Managers agree to logon asynchro-nously and give comments whenasked

• Managers agree to NOT use tool



Table 3. Continued


What Knowledge is Captured?


• Team member pointsout to team thatinformation in publicnotebook could besubpoenaed

• Team member voicesunhappiness when aconversation he hadis entered inNotebook

• Transience andspeed of knowledgegeneration makescapture of mostknowledge futile

SPIRIT: Capture all knowledgeFEATURES• Use templates to capture decision

rationale and conversations

SPIRIT: Capture some knowledgeFEATURES:• Text boxes for commenting rarely used• Templates for other than meeting minutes

not used• Repository only includes drawings,

analyses and minutes

Group Structure

• All knowledge gets captured • Post decision rationale and conversationsonly when explicitly requested

• Post analysis results and drawings atbeginning of meeting but withoutannotations

• Maintain private notebook


• Managers agree to use and postfeedback

• Only formal reviews captured• Managers don’t post feedback

How is Knowledge Captured?

Technology Structure• Pace of idea

generation was fasterthan ability to sketchnew ideas—brain-storming sessionstopped

• Complex taskrequiring interactiveidea generation andevaluation—asynch-ronous brainstormingunproductive

• Managers aren’tsufficiently know-ledgeable about toolto post comments

• Team members toobusy to logon everytime they think of aproject idea

SPIRIT: Continuously by everyoneFEATURES:• Only one author can edit entry at a

time, encouraging asynchronous use• Create sketches via drawing palette

SPIRIT: Capture only when designatedFEATURES:• Put lock on entry creation to avoid

overwriting during synchronousbrainstorming

• Scan in drawings before teleconferences• Use drawing palette to highlight, not

sketch

Group Structure

• Members agree to try brainstormingasynchronously

• Members agree to capture knowledgecontinuously as they think of ideasand comments

• Synchronous brainstorming done virtuallyaided by common-language metaphors

• No unplanned knowledge capture, i.e.,one member agrees to capture minutes,another agrees to scan in drawings


• Managers agree to use tool to reviewteam’s progress

• Manages agree to NOT use tool



Table 3. Continued


What Knowledge is Shared?


• Too many notifica-tions during brain-storming

• Knowledge toocontext-specific to beunderstood withoutverbal discussions

• Most entries givenkeyword “design,”negating value ofkeywords

• Entries too transientto create motivationto use keywords andreference links

SPIRIT: All knowledge sharedasynchronouslyFEATURES:• History of entries specified through

“precedent” reference-linking• Three keywords (maximum)/entry

allowed• Automatic e-mail notification• Private-access Notebooks can be

created

SPIRIT: All knowledge in Notebook shared,primarily during synchronousteleconferencesFEATURES:• Reference-linking not used• Keywords minimally used• E-mail notification turned off• Private-access Notebooks not created

Group Structure

• All knowledge will be shared witheveryone

• Maintain private notebook offline• Everything in Internet Notebook is

information team agrees to share• Share only what is essential for others to

see: drawings, analysis, and minutes


• Managers agree to have theirknowledge shared

• Managers not using tool so revert toneed-to-know

What Helps Knowledge Sharing?


• Keywords and linksof little value infinding entries

• Group memoryallows for quick recallwithout Booleansearches

• Reference links topast entries notunderstood by others

SPIRIT: Do own searches and makelinks between entries explicit to helpsharingFEATURES:• Entries easily navigated by date,

keywords, or reference links• Use powerful search tools (e.g.,

reference link network, Booleansearches of multiple keywords)

• No print feature

SPIRIT: Do own simple searches ofindividual entries but don’t make linksexplicitFEATURES:• Only simplistic navigation tools used (date

and author)• Powerful search tools not used• Print feature added

Group Structure

• Do own searches and make ownexplicit links between entries

• Search based on keywords or links

• I can’t find entry, ask someone inmeetings

• Search based on entry number, author, ordate


• Use keyword “Mgmt_Review” to labelmanagement-oriented entries

• Managers agree to search for entrieswith “Mgmt_Review” keyword

• Since tool not used, managers revertback to using others to find and shareknowledge



Table 3. Continued


Who Participates in Decision Making?


• Posting all designideas and analysesresults in Notebookforced lead engineerto discuss eachdesign idea andanalysis result withteam

• Negative manage-ment review forcedteam to reconsiderhow they wereinteracting withmanagement

SPIRIT: Adopt hierarchical structureFEATURES:• No decision-making aids

SPIRIT: Facilitate participationFEATURES:• Common repository to see what all team

members do

Group Structure

• Hierarchical with lead engineermaking all design decisions

• Participative discussions about eachother’s disciplines

• Joint decision making on design


• Hierarchical with team leaderresponsible for all communication tomanagers

• Managers interface not with team lead butspecialists

Questioning Technical Requirements for Decision Making


• The Project Alterna-tives Matrix entryexposes interdepen-dencies amongdisciplines

• Specialists indicatean interest in under-standing fundamen-tals of otherdisciplines

• Negativemanagement reviewmakes explicit theparadox of justifying a non-traditionaldesign based ontraditional analysismethods

SPIRIT: Adopt existing practiceFEATURES:• No aids to challenge technical require-

ments• Use external application viewers to

share documents so don’t needcommon analysis models

SPIRIT: Make assumptions explicitFEATURES:• Common repository available to all makes

others’ assumptions and the implicationof management’s technical requirements

Group Structure

• Specialists implicitly know owntechnical requirements

• Technical requirements unchallenged

• Assumptions underlying entriesquestioned during teleconferences

• Questioning causes technicalrequirements to be challenged


• Technical requirements imposed bymanagement unchallenged by team

• Managers reconsider technicalrequirements



Team members onlyuse tool synchronously

to capture and find someknowledge andmake decisions

participatively withinnegotiated constraints

Restrictaccess toteam only

Team usescommon tool

Use notool

Sync

hron

ous

primar

ily

Sync

hron

ous

only

Not u

sing t

ool s

o

sync

hron

ous o

nly

Facilitate Participation

Participative

Expert based decisionmaking between team and

management

TECHNOLOGY

GROUP

ORGANIZATION ENVT

Only formal review minutes

Some explicit knowledgeonly

By others

Capture only when designated

Everythingin the

notebookAll in

notebooksharedduring

teleconferences

Do simple

searchesalone;

use othersto findlinks

Do ownsimple

searches butnot links

III.

KN

OW

LED

GE

SHAR

ING


I. ACC

ES S TO S A M

E CO

MM

UN

ICAT IO

NTO

OL

IV. DECISION MAKING

a) W

hat h

elps

sha

ring?

B) W

hat g

ets

shar

ed?


a) Who?

b) When?

Captureonly when designated

Explicitknowledge

� + �

�

�

�

�

�

�

�

�

� �

��

�

�

�

�

+

+

+

+

++ +

+

+

+

+

+

Make assumptions explicit

Challenge technicalreqs

Technical reqs challenged in F2F discussions

a) Participation

b)Questioning techn reqs

�+

�

+

+�

+�

�

�

Use others

Need-to-knowbasis

�

LEGEND

= change from T2

+ = change from T1

Summary ofT3 Appropriations

Team members onlyuse tool synchronously

to capture and find someknowledge andmake decisions

participatively withinnegotiated constraints

Restrictaccess toteam only

Team usescommon tool

Use notool

Sync

hron

ous

primar

ily

Sync

hron

ous

only

Not u

sing t

ool s

o

sync

hron

ous o

nly

Facilitate Participation

Participative

Expert based decisionmaking between team and

management

TECHNOLOGY

GROUP

ORGANIZATION ENVT

Only formal review minutes

Some explicit knowledgeonly

By others

Capture only when designated

Everythingin the

notebookAll in

notebooksharedduring

teleconferences

Do simple

searchesalone;

use othersto findlinks

Do ownsimple

searches butnot links

III.

KN

OW

LED

GE

SHAR

ING


I. ACC

ES S TO S A M

E CO

MM

UN

ICAT IO

NTO

OL

IV. DECISION MAKING

a) W

hat h

elps

sha

ring?

B) W

hat g

ets

shar

ed?


a) Who?

b) When?

Captureonly when designated

Explicitknowledge

� + �

�

�

�

�

�

�

�

�

� �

��

�

�

�

�

+

+

+

+

++ +

+

+

+

+

+

Make assumptions explicit

Challenge technicalreqs

Technical reqs challenged in F2F discussions

a) Participation

b)Questioning techn reqs

�+

�

+

+�

+�

�

�

Use others

Need-to-knowbasis

�

LEGEND

= change from T2

+ = change from T1


�

LEGEND

= change from T2

+ = change from T1


Figure 4. Appropriated Structures at T3

each structure at T3 represents no change (nosymbol), a change from the structure at T2 (ablack triangle), or a change from the pre-existingstructure at T1 (“+”). In the text below, each of theT3 structures and the appropriations required toachieve them are described. In addition, thesecond column of Table 3 presents the data onwhich the interpretations about appropriationswere made.

Appropriated Technology at T3During the course of the project, the CT was itselfmodified in significant ways. Some of thismodification was in the addition of new features,some in the decision to not use certain features,and some in the intent to which the features wereused, i.e., the spirit or norms of use. Assummarized using triangles in the technology

circle in Figure 4, changes were observed in allappropriation topics.

With regard to access to the tool, accessprivileges were initially set to allow bothmanagement and team members access to thetool and members were expected to use externalapplication viewers (called “Hot Links”) to sharetheir documents. Team members instead even-tually limited access to only team members,deciding to exclude management. In addition,team members pushed the developer to create ascreen capture feature, which eliminated the needfor external application viewers. Team membersalso requested, and received, a feature thatlocked entry creation so that their entries wouldnot be overwritten during synchronous brain-storming. Finally, the single most significantchange was to couple virtually all use of the tool



with a synchronous “meet-me” telephone con-ference. By the end of the 40 weeks of theproject, 89 such audio/electronic virtual meetingshad been held.

The technology was appropriated to facilitateknowledge capture as well. Although there weretemplates to capture decision rationale and textboxes for commenting on drawings, the decisionrationale and comments were rarely captured,leaving these features mostly unused. In the end,the over 1,000 entries contained only explicitknowledge (such as formal drawings, analyses,and minutes), rather than tacit knowledge (suchas conversations, assumptions, or interpretations).This knowledge was also not captured in the wayexpected: the audio portion of the virtual meetingsbecame essential to interpreting entries, andknowledge was not captured continuously as ithappened (such as by using the drawing palette tocreate sketches), but rather input as entries rightbefore meetings (such as when drawings werescanned in) and then the drawing palette wasused to highlight points being discussed duringthe meeting.

The appropriations for knowledge sharing alsochanged. For example, entries were generally notcreated with the use of keywords, despite theinitial protocol that entries would be coupled withat least three keywords. Reference links to similarprevious entries were sparse. “Private” entries forprivately accessible notebooks, using person-alized keywords, were not created. Becausekeywords and reference links were not generallyused, only the simplest navigation and searchtools (such as sort by latest date or author) wereused. Finally, the team requested and received aprint feature so they could examine and sharedrawings off-line.

Even the final appropriation topic of decisionmaking experienced changes in technology.Although the spirit of the technology was to adoptthe existing hierarchical structure, the most centralfeature of the technology—a common repositoryof all critical knowledge available to all on theteam—resulted in everyone on the team askingmany more questions about each others’ ideas,drawings, and analysis results than in previousdevelopment efforts. This resulted, as will beexplained in the section below, in a technology

spirit that fostered more participation in the designprocess than the traditional hierarchical structureallowed.

In sum, as indicated by the triangle symbols inFigure 4, the final technology structure that theteam adhered to required that the team changeappropriations that had been agreed to at theoutset of the project for all eight appropriationtopics.

Appropriated Group Structure at T3Initially, the Internet Notebook was intended toserve as a knowledge repository only, with teammembers continuing to use their own geometricmodels for the specialized analyses each per-formed. Over time, however, members decided touse the repository to house a single analyticgeometric model from which the specialists couldperform their analysis. This was unprecedented,because it meant that assumptions about para-meters needed to be adopted across disciplines.

The availability of a common tool and commonanalysis model to share knowledge meant that, inprinciple, the lead engineer’s role as informationgatekeeper could be bypassed. This was, in fact,what happened; in addition, the lead engineer’srole expanded and became more ambiguous.From being the hub in a communication wheel, henow had to get everybody to explain their inputs toothers and at the same time negotiate witheveryone publicly on which solution to adopt.Initially, the lead engineer found this role changevery discomforting, complaining on more than oneoccasion about how “I’ve never seen a rocketdesigned by committee before.” Over time,however, as it became increasingly clear that thelead engineer was unable to formulate a solutionthat would meet management requirements, theincreased participation allowed the generation ofmore innovative solutions and an increasedunderstanding of the solution by all team mem-bers. While the lead engineer reported in the finalinterview that he missed being in control (andseveral others commented that greater leadercontrol might have been beneficial), he alsorecognized that his multiple time commitmentswould not have allowed him to simultaneously stayin control and generate the needed solution.



With this shift from a hierarchical to a participativedecision making structure, the role of each teammember shifted. While each specialist continuedmaking their independent technical analyses,each member also began to engage in the designprocess more proactively. For example, when anentry of an analysis result was posted, specialistsin other areas would not only query the specialistto understand what the results meant, but mighteven suggest alternative analytic interpretations ofthe results or question underlying assumptions ofthe analysis method. This led to questioning whathad initially been accepted as management-imposed technical requirements for the design, aprocess that, in the end, led to a breakthroughsolution.

While the tool housed more knowledge about theteam’s progress than had ever been capturedbefore (with over 1,000 entries), the group learnedduring the project that they could not capture allthe knowledge. Three weeks into the project, forexample, a comment by the lead engineerindicated that he had had a “private” conversationwith a team member. After much discussionabout this conversation (e.g., what should bemade public about it, how it violated the protocolfor no face-to-face meetings, etc.), the team finallydecided to sanction face-to-face meetings as longas the results of the meeting were documentedlater on. The team also found it increasinglyburdensome to try and document everythingpublicly. So, instead of voluntarily documentingconversations and informal and tacit knowledge,team members began to only document thisimplicit knowledge when explicitly asked to do so,such as by other members during a virtualmeeting. This often meant that virtual meetingswere spent describing this tacit knowledge byexplaining assumptions and context for eachentry. Moreover, the team members began toaccept designations or assignments as to whatthey were responsible for documenting (onemember documented minutes, another scanned indrawings, etc.). In this way, it was no longer thecase that everyone was responsible for docu-menting everything and that everything was beingdocumented. The team members came to viewthe Internet Notebook, then, no longer as arepository of all knowledge, but a repository of theknowledge that the team had decided to enter.They still adhered to sharing all knowledge that

was in the Notebook with all members of theteam; just not sharing all knowledge.

A final significant appropriation had to do withbrainstorming. At the outset of the project, teammembers were only familiar with the type ofintensive, collocated brainstorming that formed theessence of their creative work, in which membersconfront, “read their eyeballs,” gesture on boards,grab the marker, interrupt, and generally createnew ideas through collective physical as well asintellectual actions. While the members hadagreed to try brainstorming asynchronously (e.g.,by members creating entries and others com-menting on them), over time, it became apparentthat members were able to brainstorm virtually,but not asynchronously.

Appropriated OrganizationalEnvironment at T3For the executive management at RocketCo, theagreements they made for how they would interactwith the virtual technology-enabled team to meetthe needs of the team were changed in all eightappropriation topics. Comparing Figures 3 and 4for Organizational Environment indicates that, forsix out of the eight appropriation topics, theyreverted back to their pre-existing environment.The managers didn’t access the tool, they didn’tshare their knowledge by commenting on entries,they did not use the Notebook to search forrelevant team knowledge, they did not publiclyshare informal knowledge, they did not workasynchronously with the team by typing in entries,and they did not personally engage in capturingtheir own knowledge—all practices they hadpromised to change from the way they normallyworked.

In two areas, there were changes that reflectednot a reversion, but a new structure. First, asteam members began to use the entries andteleconferences to elicit and question technicalassumptions underlying the proposed designs, theteam developed a consensus about the impossi-bility of certain technical requirements that hadbeen initially imposed by management. This con-sensus gave individual team members enoughconfidence to challenge management. Eventually,these challenges led several managers to recon-sider technical requirements previously “written in



stone” for all projects. Second, the participativenature of the team called the traditional hier-archical relationship with management intoquestion. Since the most appropriate people torenegotiate technical requirements withmanagement were disciplinary experts, limitinginteractions between the team and managementto the project leader alone no longer made sense.In fact, one of the team’s virtual meetings explicitlyfocused on this issue of who should be speakingwith the various executive managers. The teamdecided to pair up disciplinary specialists andmanagers and have these pairs resolve disagree-ments about acceptable technical requirementsand solutions for the team.

Result 4: What Were the Reasonsfor Appropriations?

To find the reasons for each of the appropriatedstructure, we noted events that immediatelypreceded the change to the structure. Forexample, for the change in access to the InternetNotebook from everyone (including managers) toteam members only (excluding managers), theimmediately preceding event was that severalmanagers had logged in and complained that theentries they reviewed lacked the quality ofanalysis they expected. In addition to noting pre-ceding events, final interviews asked teammembers to describe their views as to why thechanges occurred. The last column of Table 3presents the list of reasons accompanying eachchange in each structure. Due to space con-straints, we have focused our discussion belownot on elaborating each reason but on describingthe patterns across the complete set of reasonswe identified.

Table 3 lists over two dozen reasons for changesin structures. These reasons for changes led usto recognize that these reasons were bestcharacterized as “discrepant events,” defined asan event that explicitly called into question anexisting structure. A review of the events indi-cated that they could not be universally attri-butable to actions of management, limitations oftechnology, or inability of the group process. Noris there one kind of event that occurred that hadreverberating impacts on all other events and

structures. Nor did they occur as part of a smallset of discontinuities. Nor were the eventsseemingly rooted in the pre-existing structures;that is, an event didn’t occur because it wassomehow “destined” to occur, as whenmanagement pre-ordains that collocation isrequired to make decisions even though manage-ment may claim to try an alternative approach.Instead of finding a single way of characterizingthe events, we instead found that the discrepantevents seem to represent four recurring themesthat the team struggled to resolve throughout thelife of the project:

(1) how to overcome barriers to addingknowledge to a public repository (e.g., beingtoo busy, knowledge transience, and aninability to clearly distinguish a priori betweenpublic versus private knowledge);

(2) what are the expectations for acting onshared information (as, for example, whenmanagement didn’t like what they saw in theNotebook, they told the team, resulting in theteam being demoralized and closing downaccess by the managers to the CT; wheninformation about designs was openly shared,the lead engineer’s job changed drama-tically);

(3) how to overcome barriers to the use of searchtools (e.g., lack of motivation to adhere todiscipline for organizing knowledge; lack ofproof that search tools are faster than people-connections); and

(4) how to share knowledge without the benefit ofcollocation cues (e.g., use of metaphors,screen capturing from past entries).

For example, there were many discrepant eventsthat reminded the team throughout the project ofthe barriers to adding knowledge to a repository.These were only discrepant events because theteam at T1 had optimistically envisioned aknowledge base containing all knowledge usefulto everyone on the current as well as futuredesign teams. Once the team scaled back fromthis vision and accepted a much more limitedknowledge base, the discrepant events ceased.



Table 4. Project Outcomes by Team Members and Managers

Outcomes Core Team (N = 8) Managers (N = 7)

Meeting Design Target for Mean S.D. Mean S.D.

Manufacturing cost 6.3 0.4 5.3 1.5

Innovation in use of collaborativetechnologies

6.0 0.9 5.3 1.1

Innovation in injector design 6.0 1.0 5.3 1.1

Performance 5.3 1.2 4.7 1.4

Performance scalability 4.3 1.0 2.7 1.1

Manufacturing scalability 3.4 1.2 3.4 1.5

Note: “1" = Very ineffective to “7" = Very effective

These discrepant events, then, were not a func-tion of pre-existing structures, but rather a functionof ideals and expectations that the team had foritself. These expectations couldn’t be fulfilled, notbecause any particular structure was a barrier, butbecause the existing state-of-the-art in thecombination of technology, organizational environ-ment, and group processes was insufficient. Forexample, while voice recognition systems couldhave captured all oral communication and video-taping could have captured the physical cues(actions actually tried by the team), the teammembers felt that it was often what was not saidand the implicit conclusions drawn at the end of aparticular design session that were more impor-tant to capture than each utterance. Yet, elicitingthese inferences was difficult for the membersbecause it might only be in later conversationsthat they realized they had made an inference.Thus, the discrepant events associated with thetheme of capturing knowledge that the teamencountered were created largely by inadequatetechnology-organization-group process solutionsavailable to them for capturing the right knowledgeat the right time in the right amount—and notsimply because of recalcitrant or non-malleablestructures thrust upon them.

Result 5: How Successful Wasthe Team?

By the end of the 10 month project, the team hadgenerated over 20 different design concepts, withthe final design concept passing the formalmanagement review. The seven executives andeight design team members were surveyed at theend of the project and asked to rate the degree towhich the team met its programmatic and tech-nical objectives. As Table 4 indicates, all felt thatthe team had met design and innovation targets,except for scalability.

A final data point indicating that the project was asuccess was that the managers approved thedesign for the next step in the developmentprocess: a cold-flow test assessing the validity ofthe analytic assumptions of liquid flow through theparts. According to one executive manager,

The team succeeded at designing athrust chamber for a new rocket enginewith only 6 parts instead of the traditionalhundreds, with a predicted quality ratingof 9 sigma (less than 1 failure out of 10billion) instead of the traditional 2 to 4



Sigma, at a first unit cost of $50,000instead of millions, and at a predictedproduction cost of $35,000 instead ofmillions. The team was able to achieveall of this with no member serving morethan 15% of his time, within the develop-ment budget, with total engineering hours10 times less than traditional teams,using a new collaborative technology withseveral partners having no history ofworking together. The team membersachieved this success through colla-boration.

The team received an award for outstandingachievement from RocketCo Senior Management.

Summary of Results

The findings can be summarized as follows:

(1) The team experienced significant misalign-ments among pre-existing structures at T1.

(2) At T2, to resolve the misalignments, the teammodified the organizational environment andgroup structure, leaving technology structureintact.

(3) Discrepant events arose as the team tried toperform with the new structures.

(4) Discrepant events resulted in further changesto the structures at T3.

(5) The final structures at T3 represented signi-ficant changes to the T2 structures.

(6) The final organizational environment revertedback to its T1 state while the technology andgroup structures represented emergent struc-tures different from both those at T1 and T2.

Answering Our Research Questions

What, then, are the answers to the four researchquestions asked at the outset?

(1) Can the workgroup adapt any or allstructures, or does it primarily try to adapt tothe technology’s initial spirit? The workgroupinitially changed the organizational environ-ment and group structures but left thetechnology’s spirit intact, as DeSanctis andPoole (1994) would predict. However, overtime, as the team was confronted withdiscrepant events that indicated that theycould not leave the technology’s spirit intact iftheir performance goals were to be achieved,the team also changed the technology spirit.Thus, we found that, when a workgroup isallowed to modify its structures, it is possiblethat all structures may be changed.

(2) Do pre-existing structures constrain theworkgroup’s adaptation process, even whenthese structures can be changed? We foundthat the notion of pre-existing structuresconstraining the adaptation process was toosimple, since the notion does not handle thecomplexity of what “constraint” means, as wellas what “structures” mean. We found that,because the workgroup was allowed tochange the structures, the structures were infact changed; so, in this regard, they were notconstraining. Moreover, we found that thepre-existing structures did not seem to predictthe discrepant events; pre-existing structureswere, therefore, also not constraining in thisregard.

However, we did find that organizationalenvironment reverted back to its pre-existingstructure, despite being modified at the outsetby the team. Does this mean that the organi-zational environment is constraining? If con-straining means that it constrained the otherstructures, the answer appears to be no, asthe discrepant events that caused changes intechnology and group structures were notnecessarily due to organizational environ-ment. If, however, constraining means thatorganizational environment was not malleablein the long run, that appears to be the case.

Finally, we found that changes in structurewere more attributable to discrepant eventsthan to malleability of the structures them-selves. For example, when the team stopped



the first virtual brainstorming session, they didnot blame the technology, organization, orgroup but instead realized that the problemwas trying to capture everything they weredoing during the brainstorming. As a result,the team continued electronic brainstormingbut revised the protocol devised at T2. Thus,a change in technology structure came aboutbecause of the discrepant event of trying tocapture too much information, not because ofa structural constraint. Where do thesediscrepant events come from? We argue thatthey are derived from the lack of goodsolutions to difficult electronic collaborationproblems, not (necessarily) from theunwillingness of recalcitrant structures tochange. Deciding what to add to a publicknowledge repository is a difficult question,even when structures are malleable.

Thus, to answer the second researchquestion: some pre-existing structures maynot be malleable in the long term (as was thecase with organizational environment), butthis lack of malleability may not affect how therest of the structures are adapted. Instead,the adaptation process is more directlyaffected by discrepant events which mayarise not because of structural constraints butbecause of the lack of available technology,organizational, and group solutions to solve adifficult coordination problem. Both malle-ability and discrepancy seem necessary, butnot sufficient, except in their interaction.

(3) After the initial adaptation to achievealignment, does the workgroup experiencethe need for further adaptations? We clearlyfound the need for further adaptation, but thisneed for adaptation did not arise because ofan initial misalignment, as Leonard-Barton(1988) would have predicted, with misalign-ments recognized and gradually resolved. Instead, at the outset, the team had theopportunity to modify its structures so thatthey could achieve what they thought wouldbe perfect alignment when they began theirwork, so further adaptation would not benecessary. In a sense, the team achievedwithin the first two weeks a set of alignedstructures for their work process. As the work

process unfolded, however, the team becameconfronted with discrepant events informingthem that the “perfectly aligned” structuresthey had created were not working—notbecause they were necessarily misaligned,but because the emergent task demanded adifferent set of aligned structures. The team,over the course of the remaining 38 weeks,tried to find ways to create a new structurethat would lead to successful performance.Thus, the process was not one of initialmisalignment gradually reduced to alignmentand successful performance, but of initialmisalignment, immediately reduced to (pre-sumed) alignment, followed by discrepantevents creating modifications to structuresthat created new misalignments, followed byfurther changes to structures to reducemisalignments, etc. To answer this thirdresearch question, then: in the context ofunpredictable work environments, having theopportunity to modify existing structures is notenough to eliminate the need for furtheradaptation. Indeed, we can imagine thatsome groups may never converge on a set ofrevised adaptations, spending most of theireffort in adjusting, and finally achieving limbo,failure, or dissolution.

(4) What is the nature of these adaptations: Arethey discontinuous responding to windows ofopportunity or are they continuous, graduallyclosing misalignments? As indicated above,we found that the Leonard-Barton (1988)model of a gradual reduction of misalignmentwas too simple for the SLICE team. We alsofound, however, that the Tyre and Orlikowski(1994) model of discontinuous adaptationsresponding to windows of opportunities didnot fit well. These discrepant events did notoccur in batches; they occurred sporadically,individually, throughout the course of theproject. Nor were the events perceived by theteam members as opportunities; they weregenerally seen as problems that aroseunforeseen and unwelcome. Finally, theevents were not windows in the sense thatthere was not a short, fairly clearly boundedamount of time in which to resolve thediscrepancy; for some of these events, it tookthe team many weeks to identify a new



structure that would resolve the discrepancywhile, for other events, resolution came inminutes. The nature of the adaptation pro-cess as applied to SLICE, then, was one ofadaptations yielding increased alignment,followed by an almost continuous array ofdiscrepant events indicating that newstructures were needed.

Why did Tyre and Orlikowski find discon-tinuous change characterized by relativelylittle change after the initial episodes, and wefound sporadic and ongoing change? Onepossible explanation might be found in thenature of the cases. Tyre and Orlikowskiargue that the costs to change were so highin their cases that, once the initial adaptationwas accomplished, further adaptation wasdiscouraged. For example, in one of theircases, the pressure for production was citedas the reason why there was little energy tosolve new problems, thus discouraging addi-tional adaptations. In the SLICE team case,however, the costs of not adapting were infact much larger than the costs of adapting.We believe that the reason for this was thatthe technology was not simply a complementto the work of the team (e.g., work-aroundswere not possible; Gasser 1986) and thatimpediments to the use of the technologywere not merely annoyances. That is, unlikeone of Tyre and Orlikowski’s cases in which“changes could be done in their leisure time”(p. 106), the technology was critical to theirwork, with impediments having direct pro-ductivity implications. Thus, we believe thatadaptations may become discontinuous whenthe costs to change are very high, and whenthe technology is large or complex; but whenthe costs to not adapt are higher, and whenthe technology is more malleable, the adapta-tions may become ongoing, if not continuous.

Discussion

Limitations

There are, of course, many methodologicallimitations of this study. As a small-sample single

case study, generalizability cannot be assessed,except as the findings conform to findings alreadypublished in the literature or as the theoreticalsuggestions prove useful in later studies. Thestrength of the study, however, lies in thecloseness and length of time in which weobserved the team’s behaviors. In addition, wehave no reason to believe that the subjects wereinherently different from other subjects we mightencounter in virtual teams: they were all success-ful professionals in their own organizations whoenjoy the challenges that virtual teaming brings.However, we do suspect that the task that theteam performed—of inherently unpredictable workon a highly innovative design—is different frommany previous studies of virtual teams. Therefore,more case studies are needed to test thegeneralizability of the team’s behaviors to othercases of unpredictable and highly innovative work.In anticipation of that work, there is much we canlearn today from the behaviors that the teamadopted.

Theoretical Implications

Together, these findings suggest a revised orextended model of structural adaptation, one thatintegrates our findings with those of DeSanctisand Poole (1994), Leonard-Barton (1988), andTyre and Orlikowski (1994). This model, pre-sented in Figure 5, adopts the distinctions amongstructures suggested by DeSanctis and Poole astechnology, group, and organizational environ-ment. We also concur with them that appro-priation moves lead to decision processes, whichin turn lead to (ideally, but not always, positive)outcomes. In addition, however, we suggest thatthe effect of pre-existing structures on appro-priation moves is not a direct one, as DeSanctisand Poole suggest, but is instead mediated bythree factors: the degree of misalignment (fromLeonard-Barton 1988), the malleability of thestructures (Johnson and Rice 1987), and theoccurrence of discrepant events (from Tyre andOrlikowski 1994). However, unlike the suggestionby Tyre and Orlikowski, these discrepant eventsare not necessarily discontinuous but rather occurpotentially continuously over the life of an adapta-tion process (depending on the size, cost, andtime frame). Moreover, the discrepant events do



MISALIGNMENT ALIGNMENT

PREEXISTINGSTRUCTURES

EMERGENTSTRUCTURES

Sources of Structure:TECHNOLOGY

GROUPORGANIZATION ENVRONMENT

DISCONFIRMINGEVENTS

APPROPRIATIONMOVES and

FAITHFULNESS

DECISIONPROCESS POSITIVE

OUTCOMES

malleability

malleability



EMERGENTSTRUCTURES



DISCONFIRMINGEVENTS


FAITHFULNESS


OUTCOMES



EMERGENTSTRUCTURES



DISCONFIRMINGEVENTS


FAITHFULNESS


OUTCOMES

malleability

malleability

Figure 5. Model of Adaptation Process

not necessarily result from pre-existing structuresas DeSanctis and Poole would suggest, but mayinstead arise from emerging events. Thediscrepant events can lead to increased misalign-ments (as suggested by Johnson and Rice),instead of the necessarily gradual reduction inmisalignments that Leonard-Barton suggests. Weconcur with all the above models that emergentstructures are likely to occur, but we suggest thatthese emergent structures may themselves createnew discrepant events. We concur with the Tyreand Orlikowski and the Leonard-Barton models bysuggesting that high-quality decision processesalone do not account for successful adaptation;alignment is required as well. Finally, we believethat any of these structures—technology, organi-zational environment, or group—are inherentlyable to change in this structuration process; oneshould not be seen as necessarily any moreconstraining than another, although in a particularcontext any particular structure’s malleability maybe restricted.

Thus, by examining a field setting in whichmalleability of all existing structures washeightened, we found that differences betweenthe models could be resolved. The differencesbetween Leonard-Barton and DeSanctis andPoole about whether certain structures are moremalleable can be resolved with the conclusion thatno structure is inherently more malleable than anyother structure; the constraints typically attri-butable to non-malleable structures are morelikely due to the field setting than inherent in thenature of the structure. The differences betweenDeSanctis and Poole and Tyre and Orlikowskiabout the size of adaptations can be resolved withthe conclusion that the number or type of mis-alignments is less relevant to adaptation successthan the resolution of discrepant events. Finally,the difference between Tyre and Orlikowski andLeonard-Barton about the continuity of adapta-tions can be resolved with our conclusion thatadaptations are in response to the scale andfrequency of discrepant events which will occurthroughout a team’s lifecycle.



The model we pose for resolving differencesbetween previous models offers several sugges-tions for further research. First, the appearance ofsignificantly greater constraints by technologyspirit (or any other structure) in prior research maybe an artifact of confounding malleability ofstructure with type of structure, time scope of thestudy, size and complexity of the technology, andcosts and benefits of adaptation. Thus specificstudies, as well as comparisons and reviews ofstructuration results, should consider the influencethese factors may have in shaping the results.

Second, researchers should not assume that anyparticular pre-existing structure is necessarilyfixed (though in practice certainly some of themare highly stable), as Griffith (1999) arguesconcerning technology features. Studies ofadaptation processes should attempt to identify allsalient misalignments and associated discrepantevents and appropriations. Thus, an “ideal”technology implementation should not be definedas one in which misalignments do not occur,because adaptations large and small will beneeded, nor as one in which users faithfully hewto the technology’s spirit; instead, an idealtechnology implementation should be definedbased on the ability of the team to resolve its ownmisalignments and the range of structuresavailable to appropriate.

Third, this study has provided some insights anda model that may help to incorporate research onvirtual teams into the mainstream of informationtechnology research by looking at virtual teams asa new organization with malleable structuralconditions. Conversely, it also suggests thatstudies of collaborative technology must be wellembedded within organizational and work groupcontexts.

Practical Implications

First, given the range and number of adaptationsthat occurred in the SLICE project, managersresponsible for implementing new technologymust allow adaptation in all three structures—technology, organizational environment, and workgroup structures—neither assuming that initialappropriations will be sufficient, nor that the need

for adaptation can be avoided. This means that akey function of team leaders should be to makethoughtful choices about which structures shouldbe malleable and which should be adapted, ratherthan to assume that adaptations are constrainedby whichever structures appear to be most malle-able. In addition, the critical role of discrepantevents in reevaluating adaptation choicessuggests that managers must create an environ-ment in which discrepant events are openlydiscussed as a catalyst for recurring adaptations.Thus, managers may want to suggest adaptationsnot always because they are convinced a priorithat the adaptations are correct, but because theadaptations may create discrepant events which,if openly discussed, provide insight for additionaladaptations. The numerous adaptations suggestthat designers of collaborative technologies (CTs)should not over-emphasize rigid technologyfeatures that cannot be adapted to changingneeds of users. In accord with Griffith, wesuggest that designers and implementers shouldbetter understand users’ perceptions of features,rather than presume that their conception of thetechnology’s spirit, and the associated features,are “fixed” structures. For example, CTs shouldbe designed to allow changes in any initialcoordination protocols as the team’s relationshipsand understandings evolve. The CT should allowfor just-in-time learning about the CT’s features.Further, it is clear from the SLICE case that theteam members gained a great deal from theirteleconferences. This suggests that CTs shouldbe designed with an expectation that they will becoupled with informal and oral forms of com-munication that are not necessarily face-to-face(Krauss and Fussell 1990; Kraut and Streeter1995; Schrage 1990; Whittaker and O’Conaill1997).

The SLICE case also identified four sources ofdiscrepant events that virtual teams will en-counter, including barriers to adding knowledge,clarifying expectations for acting on sharedinformation, barriers to the use of search tools,and ways to share knowledge without the benefitof collocation cues. While we are not suggestingthat virtual teams can resolve these issues at theoutset, we are suggesting that CT developers andvirtual team managers should attend to theseissues in the preparation and support of the virtual



team. Finally, our study suggests that virtualteams using new CTs need to allow themselvesthe flexibility to evolve their own knowledge-sharing norms over time. The team was providedsufficient autonomy to establish its own norms ofinteracting with a highly flexible technology, yetstill changed those norms later on. This is perhapsnot surprising. The workgroup literature (e.g.,Gabarro 1990; Krauss and Fussell 1990) suggeststhat the process of developing “mutual expecta-tions” is an iterative one since one’s expectationsare not clear until after some experience inworking with others and on the tasks. Thus,frequent checkpoints, lessons learned sessions,and group reflections on the process shouldprobably be the most stable norm of the team.

Acknowledgements

The authors would like to thank the followingindividuals (in alphabetical order) who generouslyoffered their time and energy throughout the datacollection for research: Steve Babcock, HollisBostick, Dave Bremmer, Hal Buddenbohm,Robert Carman, Scott Claflin, Bob Corley, DennisCoston, Linda Finley, Terry Kim, Vern Lott, DaveMatthews, and Li-Kiang Tseng. The authors wouldalso like to thank the MIS Quarterly reviewers andeditors for comments on earlier versions.

References

Alavi, M., and Carlson, P. “A Review of MISResearch and Disciplinary Development,”Journal of Management Information Systems(8:4), 1992, pp. 45-62.

Barley, S. R. “Technology as an Occasion forStructuring: Evidence from Observations of CTScanners and the Social Order of RadiologyDepartments,” Administrative Science Quarterly(31), 1986, pp. 78-108.

Bowers, J. “Making It Work: A Field Study of a‘CSCW Network’,” The Information Society(11), 1995, pp. 189-207.

Contractor, N., and Eisenberg, E. “Communica-tion Networks and New Media in Organizations,”in Organizations and Communication Techno-logy, J. Fulk and C. Steinfield (eds.), Sage,Newbury Park, CA, 1990, pp. 143-172.

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

Gabarro, J. “The Development of Working Rela-tionships,” in Intellectual Teamwork: The Socialand Technological Bases of Cooperative Work,J. Galegher, R. E. Kraut, and C. Egido (eds.),Erlbaum, Hillsdale, NJ, 1990, pp. 79-110.

Gable, G. “Integrating Case Study and SurveyResearch Methods: An Example in InformationSystems,” European Journal of InformationSystems (3:2), 1994, pp. 112-126.

Gasser, L. “The Integration of Computing andRoutine Work,” ACM Transactions on OfficeInformation Systems (4:3), 1986, pp. 205-225.

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, 1984.

Griffith, T. “Technology Features as Triggers forSensemaking,” Academy of ManagementReview (24:3), 1999, pp. 472-488.

Grudin, J. “Groupware and Social Dynamics,”Communications of the ACM (37:1), 1994, pp.93-105.

Harvey, L., and Myers, M. D. “Scholarship andPractice: The Contribution of EthnographicResearch Methods to Bridging the Gap,”Information Technology & People (8:3), 1995,pp. 13-27.

Jarvenpaa, S. L., and Ives, B. “The Global Net-work Organization of the Future: InformationManagement Opportunities and Challenges,”Journal of Management Information Systems(10:4), 1994, pp. 25-58.

Jarvenpaa, S. L., and Leidner, D. E. “Communi-cation and Trust in Global Virtual Teams,”Journal of Computer-Mediated Communication(3:4), 1998 [online journal].

Johnson, B., and Rice, R. E. Managing Organi-zational Innovation, Columbia University Press,New York, 1987.

Krauss, R., and Fussell, S. “Mutual Knowledgeand Communicative Effectiveness,” in Intellec-tual Teamwork: The Social and TechnologicalBases of Cooperative Work, J. Galegher, R. E.Kraut, and C. Egido (eds.) Erlbaum, Hillsdale,NJ, 1990, pp. 111-144.



Kraut, R., and Streeter, L. “Coordination inSoftware Development,” Communications ofthe ACM (38:3), 1995, pp. 69-81.

Lee, A. S. “Integrating Positivist and InterpretiveApproaches to Organizational Research,”Organization Science (2), 1991, pp. 342-365.

Leonard-Barton, D. “Implementation as MutualAdaptation of Technology and Organization,”Research Policy (17), 1988, pp. 251-267.

Lewis, I. M. Social Anthropology in Perspective,Cambridge University Press, Cambridge, UK,1985.

Lipnack, J., and Stamps, J. Virtual Teams, Wiley,New York, 1997.

Markus, M. L. “Electronic Mail as the Medium ofManagerial Choice,” Organization Science (5:4),1994, pp. 502-527.

McKenney, J., Zack, M., and Doherty, V. “Com-plementary Communication Media: A Compari-son of Electronic Mail and Face-to-Face Com-munication in a Programming Team,” inNetworks and Organizations, N. Nohria and R.Eccles (eds.) Harvard Business School Press,Boston, MA, 1992, pp. 262-287.

Mohrman, S., Cohen, S., and Mohrman, A. Jr.Designing Team-Based Organizations: NewForms for Knowledge Work, Jossey-Bass, SanFrancisco, 1995.

Myers, M. D. “Qualitative Research in InformationSystems,” MISQ Discovery, 1997 (http://www.misq.org/misqd961/isworld).

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

Orlikowski, W. J., and Baroudi, J. J. “StudyingInformation Technology in Organizations:Research Approaches and Assumptions,” Infor-mation Systems Research (2) 1991, pp. 1-28.

Orkiloswki, W. J., and Gash, D. “TechnologicalFrames: Making Sense of Information Tech-nology in Organizations,” ACM Transactions onInformation Systems (12), 1994, pp. 174-207.

Orlikowski, W. J., and Robey, D. “InformationTechnology and the Structuring of Organi-zations,” Information Systems Research (2),1991, pp. 143-169.

Orlikowski, W. J., and Yates, J. “Genre Reper-toire: The Structuring of Communicative inOrganizations,” Administrative ScienceQuarterly (39:4), 1994, pp. 541-574.

Orlikowski, W. J., Yates, J., Okamura, K., andFujimoto, M. “Shaping Electronic Communi-cation: The Metastructuring of Technology in theContext of Use,” Organization Science (6:4),1995, pp. 423-443.

Poole, M. S., and DeSanctis, G. “Understandingthe Use of Group Decision Support Systems:The Theory of Adaptive Structuration,” inOrganizations and Communication Technology,J. Fulk and C. Steinfield (eds.), Sage, NewburyPark, CA, 1990, pp. 173-193.

Purser, R. E., Pasmore, W. A., and Tenkasi, R. V.“The Influence of Deliberations on Learning inNew Product Development Teams,” Journal ofEngineering and Technology Management(9:1), 1992, pp. 1-14.

Rice, R. “Network Analysis and Computer-Mediated Communication Systems,” inAdvances in Social and Behavioral Sciencefrom Social Network Analysis, S. Wassermanand J. Galaskiewicz (eds.), Sage, NewburyPark, CA, 1994, pp. 167-203.

Rice, R. E., and Gattiker, U. “New Media andOrganizational Structuring of Meanings andRelations,” in New Handbook of OrganizationalCommunication, F. Jablin and L. Putnam (eds.),Sage, Newbury Park, CA, 1999.

Romano, N. Jr., Nunamaker, J., Briggs, R., andVogel, D. “Architecture, Design, and Develop-ment of An HTML/Javascript Web-Based GroupSupport System,” Journal of the AmericanSociety for Information Science (49:7), 1998,pp. 649-667.

Schrage, M. Shared Minds: The New Techno-logy of Collaboration, Random House, NewYork, 1990.

Townsend, A., DeMarie, S., and Hendrickson, A.“Virtual Teams: Technology and the Workplaceof the Future,” The Academy of ManagementExecutive (12:3), 1998, pp. 17-29.

Tyre, M. J., and Orlikowski, W. J. “Windows ofOpportunity: Temporal Patterns of Technolo-gical Adaptation In Organizations,” OrganizationScience (5:1), February, 1994, pp.98-118.

Walsham, G. “Interpretive Case Studies in ISResearch: Nature and Method”, EuropeanJournal of Information Systems (4), 1993, pp.74-81.

Warkentin, M., Sayeed, L., and Hightower, R.“Virtual Teams Versus Face-to-Face Teams:An Exploratory Study of a Web-based Con-



ference System,” Decision Sciences (28:4),1997, pp. 975-996.

Williams, F., Rice, R. E., and Rogers, E. M.Research Methods and the New Media, Sage,Newbury Park, CA, 1988.

Whittaker, S., and O’Conaill, B. “The Role ofVision in Face-To-Face and Mediated Commu-nication,” in Video-Mediated Communication, K.Finn, A. Sellen, and S. Wilbur (eds.), LawrenceEarlbaum, Mahway, NJ, 1997, pp. 23-50.

Yates, J., and Orlikowski, W. J. “Genres ofOrganizational Communication: A Structura-tional Approach to Studying Communicationand Media,” Academy of Management Review(17:2), 1992, pp. 299-326.

Yin, R. K. Case Study Research, Design andMethods (2nd ed.), Sage, Newbury Park, CA,1994.

About the Authors

Ann Majchrzak is professor of informationsystems at the Marshall School of Business at theUniversity of Southern California. She holds aPh.D. in Social Psychology from the University ofCalifornia, Los Angeles. She was previously withthe Institute of Safety and Systems Managementat the University of Southern California. Herresearch interests focus on the development ofchange plans that optimize the synergy betweencomputer-based technologies, human capabilities,organizational structure, and business strategy.She has applied her research in such industrysectors as manufacturing, assembly, and engi-neering design. She has used her research togenerate tools to help technology and organi-zational designers, including HITOP, ACTION,and TOP Modeler (www.topintegration.com). Hertools have been used in Europe, Australia, Northand South America, and with such companies asFord, Hewlett-Packard, General Motors, TexasInstruments, and Hughes. Majchrzak has servedon three National Academy of Sciences com-mittees, written seven books, including TheHuman Side of Factory Automation, has a 1996Harvard Business Review article on “Building aCollaborative Culture in Process-Centered Organi-zations,” and is the 2000 winner of SIM’s Inter-national Paper Award Competition. Her currentresearch interests include development of know-

ledge management tools and processes and thedesign of stakeholder participation processes inthe IS development.

Ronald E. Rice (B.A., Columbia University; M.A.,Ph.D., Stanford University) has co-authored or co-edited The Internet and Health Communication(2000); Accessing and Browsing Information andCommunication (2001); Public CommunicationCampaigns (1st ed., 1981; 2nd ed., 1989; 3rd ed.2000); The New Media: Communication,Research and Technology (1984); ManagingOrganizational Innovation (1987); and ResearchMethods and the New Media (1989). He hasconducted research and published widely in com-munication science, public communication cam-paigns, computer-mediated communication sys-tems, methodology, organizational and manage-ment theory, information systems, informationscience and bibliometrics, and social networks.He has served as an associate editor for HumanCommunication Research as well as for MISQuarterly. He is on the editioral boards of Journalof the American Society for Information Science,Communication Theory, Human CommunicationResearch, New Media and Society, Journal ofManagement Information Systems, and Journal ofBusiness Communication.

Arvind Malhotra is an assistant professor ofInformation Technology and e-Commerce at theKenan-Flagler Business School, University ofNorth Carolina at Chapel Hill. His current researchinterests are in the area of e-business metrics, e-commerce business models, digital supply chains,and e-service quality. He is two-time winner ofSIM International Paper award competition (1997,2000). He has presented his research at Inter-national Conference for Information Systems, SIMWorkshop, and Association for Information Sys-tems conference. He currently is the director ofthe e-Launcher, an e-business incubator at theKenan-Flagler Business School. He is also aprincipal researcher at eUNC where he hasreceived grants from Dell and MSI for research one-commerce issues. He holds a BS in Electronicsand Communication Engineering, an MS inIndustrial Engineering, and a Ph.D. in InformationSystems Management from the University ofSouthern California.



Nelson King received a B.S. from ColumbiaUniversity School of Engineering and an M.S. fromthe University of Arizona. His experience in pre-liminary design encompasses natural resource,aerospace, and information technology projects.He has published in IEEE Transactions on Engi-neering Management and Project ManagementJournal. He is a Ph.D. candidate in the Depart-ment of Industrial and Systems Engineering atUniversity of Southern California. His currentresearch examines the impact of ambiguity oninnovation in the concept development phase ofnew product development.

Sulin Ba is assistant professor of information sys-tems and the co-director of the Electronic Eco-nomy Research Program (ebizlab) at the Marshall

School of Business at the University of SouthernCalifornia. She received her Ph.D. from theUniversity of Texas at Austin. Her researchinterests include electronic commerce, knowledgemanagement, and virtual teams. Her currentprojects involve the design of trusted third partiesto help small business overcome the onlinebarriers such as security and product qualityuncertainty. Her work on the institutional setup tohelp small business survive and grow in the digitaleconomy has been used as the basis fortestimony before the House Committee on SmallBusiness. In addition, she works on designingnew market-oriented organizational coordinationmechanisms. She serves as an associate editorfor Journal of Decision Support Systems.

technology daptation the case of a omputer-supported inter ...majchrza/virtualteamsinmisq.pdf ·...

Documents