Advantages and limitations of knowledge networks as a ... - CiteSeerX

6 downloads 321120 Views 173KB Size Report
of knowledge networks as a mechanism for sustaining software process ... software company becomes the higher the demand will be for continuous process.
SUBMITTED TO “Learning Software Organizations 2006”

Advantages and limitations of knowledge networks as a mechanism for sustaining Software Process Improvement Thomas Elisberg1, Jacob Nørbjerg2, Jan Pries-Heje1 1Design of Organizational IT, IT University of Copenhagen Rued Langgaards Vej 7, DK2300 Copenhagen, Denmark 2 Department of Informatics, Copenhagen Business School, Howitzvej 60 DK-2000 Frederiksberg C, Denmark

Abstract. A knowledge network is a formalized mechanism for supporting the identification, creation and sharing of professional knowledge. In this paper we report from a case study where knowledge networks were used as a core mechanism in a software process improvement effort. In the concrete we studied two networks: one on software testing and one on user experience. Among the knowledge network members the test network is regarded as a success while the user experience network experienced the mechanism as a failure. Based on a literature survey we develop a matrix which we use to characterize the actual use of the knowledge network by the two member groups. Grounded in this analysis we identify advantages and limitations of knowledge networks as a mechanism for sustaining software process improvement.

1 Introduction Organizations involved in software process improvement (SPI) often turn to norm based frameworks such as capability maturity models, e.g. the CMMI [4], to establish meaningful standards for the systems development process. Still it will be a central issue to define and disseminate best work practices in order to honor the process requirements and capitalize on the benefits of SPI. In general, the more mature a software company becomes the higher the demand will be for continuous process evaluation and innovation. From a knowledge management (KM) perspective this implies that software organizations aiming to improve the quality of their processes and products, must define and implement an appropriate knowledge management (KM)) strategy that enables them to (1) identify software processes with an improvement potential; (2) define new processes; (3) document the new processes, and (4) disseminate new process knowledge among professional software developers in the organization. Several solutions to these problems have been suggested in general knowledge management, as well as in software process improvement research; e.g. [20] [14]. Knowledge networks are mechanisms that have gained increasing attention within the field of KM. Recognizing that not all knowledge lend itself to be captured, codified and stored [10] the social aspects of KM have become increasingly popular

2

Thomas Elisberg1, Jacob Nørbjerg2, Jan Pries-Heje1

to explore. Several research contributions report on the use of networks of practitioners to create and spread knowledge. Knowledge networks can be described as structures that seek to bring peers together that hold certain specialized skills and levels of expertise [21]. Such networks may grow out of the practice of professionals [22] or be managerially designed [18]. Using networks of people to create and maintain knowledge have been successful in some cases [20] though a potential drawback – from an organizational viewpoint – may be that they are vulnerable to staff turnover and that knowledge is difficult to disseminate beyond the network [8]. This paper analyses and compares the experiences of two knowledge networks in a software development organization engaged in SPI. The two knowledge networks discussed in this paper are part of a total of seven networks formed two years ago, inspired by the team model in the Microsoft Solution Framework [15]. The research is motivated by an internal survey conducted by the organization of the seven knowledge network that showed significant differences in performance and evaluation among the networks. This study seeks to understand why this imbalanced assessment have been reported, by analyzing and discussing the findings in relation to two important dimensions in KM research: One dimension represent the distinction between the different KM strategies of codification and personalization [9] while the other dimension distinguish between exploitation and exploration [13]. The next section briefly discusses the knowledge network idea and introduces the analytical framework. Section 3 discusses the research method, and section 4 introduces the case organization. The findings from the study are further described and analyzed in section 5. Section 6 discusses the findings and their implications for creation and management of knowledge networks in general.

2 Knowledge networks in software organizations The knowledge-based perspective on organizations is ambiguous and has many facets [1]; yet it is generally accepted that knowledge is embedded in individuals, assets and structures such as employees, procedures, policies, culture, systems and documents [7; 19]. There is, however, an ongoing discussion in extant literature, of whether knowledge is a resource that can be managed and controlled as a tangible asset [5], and enacted by means of codification [17], or on the other hand, if knowledge is an intangible asset that resides within people with no possibility for being externalized or even articulated [18]. These competing views on knowledge imply different knowledge management strategies. On the one hand a codification strategy where knowledge is externalized from individuals and made available to others through for example a knowledge repository or a documented process. An alternative strategy, personalization, is based on the view of knowledge as an intangible asset where personal relations, face-to-face contact, brainstorming and conversations are central in knowledge creation and sharing. It is, however, increasingly acknowledged, that neither of these approaches are sufficient, and that organizations must base their knowledge management strategy on a balanced combination of the approaches based on the organizational context [9; 14].

Advantages and limitations of knowledge networks as a mechanism for sustaining Software Process Improvement

3

Another important distinction in KM research concerns how organizations distinguish between the exploitation and exploration of knowledge [13]. Exploitation refers to when the organization is using what is already known. This involves identification of best practices, execution and routinization. Exploration on the other hand is when the organization focuses on discovering new knowledge. Exploration involves identification of new practices, experimentation and risk taking. Like in the case of knowledge sharing it is recommended to maintain a balance between exploitation and exploration [2; 12]. However, it has been demonstrated that it is hard for organizations to balance the two approaches; this is referred to as the exploration/exploitation trade-off [13]: If an organization concentrates on exploring new knowledge it will become harder to improve existing skills. Similarly if a company concentrates on the utilization of existing knowledge it will become harder to integrate new knowledge in the process. Figure 1 combines these two distinctions in a two by two matrix which constitute the analytical framework. Exploration

Exploitation

Codification

Enabling

Retrieving

Personalization

Expanding

Connecting

Fig. 1. The Knowledge Process Typology Matrix

The Knowledge Process Typology Matrix defines four different perspectives on the knowledge process; each rooted in a distinct KM strategy and focus: -

-

Enabling is rooted in a codification strategy where knowledge creation is defined as a formal process. Acknowledging that knowledge can reside external to humans this perspective opens up for the possibility that the exploration of new knowledge can be supported by formal organizational practices. Retrieving is based on the same premises as Enabling. Knowledge from this perspective exists independent of humans and can as such be stored and made available for others to facilitate exploitation. Connecting adopts a personalization strategy and acknowledges that knowledge resides within the individual. From this perspective exploitation is facilitated by making it possible for peers to establish relations to exchanging knowledge.

4

Thomas Elisberg1, Jacob Nørbjerg2, Jan Pries-Heje1

-

Expanding is based on the same basic principles as the Retrieving perspective; however, exploration is central. Expanding refers to the growth of the network when new nodes are introduced, hereby making it possible for individuals to establish novel relations.

The literature suggests that to maintain a healthy KM strategy an organization must balance the two dimensions in the Knowledge Process Typology Matrix in a way that consider the overall needs of the business. According to the developed framework an organization must include the Creating and Sharing as well as the Expanding and Connecting perspective to succeed in its KM initiatives.

3 Data collection and analysis The research is empirically grounded in an ongoing action research project focusing on software process improvement. This particularly study can however best be described as a case study within a research project using an action research methodology [6]. The adopted case study approach is explanatory in nature [23] and is based on a qualitative approach to data analysis [16]. The study is based on ten in-depth interviews; five from each of the two networks. Three different roles in the User Experience and Test Knowledge networks were represented: two members, to persons from leadership and a leader (see case description for details). The interviews were each scheduled to last one hour, and were all tape recorded. Subsequently the interviews were transcribed to ease the analysis of the data. The data is analyzed by plotting the interviewees’ actual use and perception of the knowledge networks into the four distinct categories defined by the Knowledge Process Typology Matrix (see figure 1). The analysis is based on statements and interpretations of the knowledge networks as a mechanism for sustaining software process improvement as perceived by the interviewees. Where appropriate the analysis is supported by direct quotations to give credibility to the analysis.

4 Case description The research site is SoftHouse, a pseudonym for a medium-sized European software house. The company, which employs more than 300 software engineers, develops and sells mission critical software products and services within the health and defense industries. Back in 1997 the company decided to engage in SPI and has since committed substantial financial and human resources to this. Today SoftHouse is certified level 5 according to the CMMI model. In SoftHouse top management has regarded effective knowledge creation and dissemination as a key enabler in the SPI effort, and has launched several initiatives, including: development of individual competencies, an expert database, and a best practice repository. The experiences with these initiatives have been mixed in nature: The developers have embraced the opportunity to acquire individual competencies,

Advantages and limitations of knowledge networks as a mechanism for sustaining Software Process Improvement

5

but the initiatives have been costly and have not contributed significantly to crossproject sharing of software process knowledge; an important prerequisite for achieving the CMMI level 3 goal of common organizational processes, and the level 4 and 5 goals of statistical process control and continuous improvement [3]. The expert database and the best practice repository have had limited effect. 4.2 The knowledge networks SoftHouse has for the past four years based their software process improvement efforts on the CMMI framework. From a KM point of view, the CMMI model relies heavily on exploiting knowledge through a codification strategy. In 2004 top management decided to launch a new concept called Knowledge Networks (KN), realizing that maturity level 5 on the CMMI model required a more systematic and organization-wide approach to KM. In SoftHouse the KNs are intended as an organizational mechanism for knowledge sharing highly inspired by the concept of Communities of Practice [11; 22]. One of the main objectives of the KNs are to represent the respective disciplines in the systems development process and to assist projects in applying the knowledge in practice. The concept as it is implemented today consists of seven networks representing specific knowledge areas inspired by the Microsoft Solution Framework [15], supplemented with a network for Architecture. Each KN is managed by a group of individuals identified as Knowledge Leadership (KL). In KL one person is appointed formal leader holding the formal responsibility for the organizational implementation. The KL is responsible for collecting and analyzing improvement proposals gathered from projects and external sources within their own specific domain. Employees are members of one or more of the KNs and are expected to attend courses, workshops and contribute with ideas and knowledge where possible. KN members cannot expect to get compensation for all their activities in the KN; e.g. and workshops outside normal work hours. Each network has around 30 members including the 4 individuals that constitute KL.

5 Analysis The two networks of interest to this study, the Test and the User Experience network were established under quite different conditions. Prior to the introduction of the network structure, the test engineers had a clearly defined and shared process, whereas the User Experience network was established as a new discipline in SoftHouse with no existing practice as a base for the network. The Test Network was from the outset able to draw on a shared practice to support exploration and exploitation of knowledge in the new network structure, but the User Experience network lacked, on the ther hand, a common understanding of the discipline they were supposed to represent. Therefore, the most important task for User Experience was to

6

Thomas Elisberg1, Jacob Nørbjerg2, Jan Pries-Heje1

establish a process for their field in SoftHouse to define the discipline in relation to the systems development process. In the following sections we will analyze each of the two networks using the Knowledge Process Typology Matrix (figure 1). 5.1 The User Experience network User Experience (UE) is a new element in SoftHouse’s software development process legitimized through the adoption of the Microsoft Solution Framework team structure. The most prominent task for UE is to safeguard the interests of the end-user. During the interview the leader of UE was quite proud of having succeeded in bringing together four very different personalities in Knowledge Leadership and make them define a common process which in the future could be used to exploit UE knowledge. Since the establishment of the UE network, howewer, time has been used to explore different methods and techniques, and the UE knowledge network has in practice mainly been an effective mechanism for exploration. The KN members of the UE network have been excluded from this process. The knowledge process is characterized as an enabling process, exclusively performed by knowledge leadership. Since UE is a new discipline, no community-of-practice existed prior to the establishment of the knowledge network. Several developers and project managers did not know what to use the UE people for and they did not know what to expect from UE. The four knowledge leads more or less constitute the professional expertise within the discipline – the rest of the members of the KN are working with UE related tasks but do not have professional expertise in the area. The four KLs are thus the only resources that can review project products and deliveries from an UE perspective. This represent a limited support of the connecting knowledge process. The limited success of rooting the UE network in a personalization strategy can be subscribed to the fact, that all the knowledge leaders are quite new to the company. None of them have the in-house network that we found in the Test network. One of the knowledge leaders said: “It is hard to identify who-knows-what in this organization”. One initiative that is planned by the UE network was the preparation of in-house training sessions and course material that aims to demonstrate the value of UE persons in projects. This initiative is targeted at project managers to emphasize the value of the UE perspective in software projects. This is clearly defined as exploitation of existing knowledge; however no seminars have, however, been completed, although the course material exists. This suggests that the retrieving knowledge process is used where the connecting process still hasn’t been carried into effect. An important part of being a Knowledge Network is to engage its members in networking activities. Every Network is required to make arrangements that members can sign up for. The leads in UE find it hard to come up with arrangements. The members are not challenged in any professional ways, though they like the arrangements from a social point of view. The expanding knowledge process is therefore not utilized in any ways. To sum up our analysis of the UE knowledge network we found that effective exploitation through processes requires a common practice among members. At this

Advantages and limitations of knowledge networks as a mechanism for sustaining Software Process Improvement

7

point in time, the UE network is characterized by having a defined process but it is not shared and adopted in the projects. 5.2 The Test Network The Test Network distinguishes itself from the UE network by being founded on an established discipline within software engineering. Compared to the UE, the Test network shares a common process that acts as a frame of reference for practice. The network has more interaction between the knowledge leaders and knowledge members, more project related activities and a more experienced leadership. Furthermore the majority of projects in SoftHouse have had dedicated test resources for some time, making testers an established group in the systems development process. Several of the interviewees questioned, however, whether the Test network had added any value to existing knowledge sharing. Most of the interviewees regarded to some degree the network as a formalization of the existing informal peer networks. The formalization of the networks has had some impact on the organization. The establishment of the Test network has resulted in a systematic scanning of the context in which the Test Network operates both internally and externally. The interviews reveal that the Test KN is entering a more formal role in the SPI work in SoftHouse. Today, new ideas and improvement proposals are evaluated in a systematic way, that provides a mechanism for collecting improvement proposal. This amounts to high support to the enabling knowledge process in the Test KN. In contrast to the UE network the Test network has formed a clear internal vision: the objective is to become the best test unit in the region in which SoftHouse operates. The means to accomplish the objective include internal education, seminars and conference participation, thus actively involving all four knowledge processes. The most prominent role of the Test KN is to act as a mediator for test knowledge and to disseminate this to the rest of the organization making extensive use of the retrieving knowledge process. The Network is, however, not utilized to solve tangible problems in practice. The network members rely on their own personal networks in this regard. The network members also point to the fact that the networks do not facilitate the exchange of experiences, neither among members, nor to the organization. Thus, the Test Network does not facilitate the connecting knowledge process, though the public arrangement organized by the UE leadership support this to a certain extend. Members of the network do however not recognize this as an important aspect. The members point to the professional groups defined by the projects as the place where knowledge is shared among peers. The expanding knowledge process is likewise supported to a great extend in these groups circumventing the Knowledge Network mechanism. People tend to use their personal contacts and the informal network instead of the formalized knowledge network. To conclude our analysis of the Test knowledge network we found that the network was based on a codification strategy and sustained to a high extend both the

8

Thomas Elisberg1, Jacob Nørbjerg2, Jan Pries-Heje1

enabling and retrieving process. However the connecting and expanding processes were not facilitated by the mechanism but were rather taken care of in settings.

6 Discussion and conclusion The knowledge networks in this study are characterized by being managerial instilled and not centered on practice, but rather on the common processes used by projects in the organization. The analysis demonstrates that Knowledge Networks as implemented in SoftHouse serve primarily as a codification mechanism supporting the enabling and retrieving knowledge processes. However, a successful implementation, as demonstrated by the two case networks, is highly dependent on a shared existing practice among peers. It was the intention that the organizational implementation of knowledge networks would sustain the creation and sharing of knowledge based on a Community of Practice approach. The analysis of the two networks demonstrates however that a personalization strategy is not supported in any way by either of the networks. Both networks have been successful from a codification perspective where little or no value has been added seen from a personalization strategy. Both KNs were perceived as less successful by their members as a mechanism for day-to-day knowledge sharing and problem solving. The difference in performance between the networks as reported in the survey seems to be related to practice. The analysis indicates that the members in the test network make use of a personalization strategy, which is established upon a shared practice. An important task for the KNs is thus to provide a reference point that defines the process. The projects are however the place where practice is exerted. The interface between KN and projects thus becomes important to balance the codification and personalization strategy. It is evident that the process view derived from years work with SPI has imbued the organization with a strong focus on processes. The knowledge networks can be perceived as formal organizational structures that are designed around this line of thinking. The Knowledge Networks hence sustains SPI from a codification strategy point of view, since they are designed around core values inherited in the process view: codification and documentation. From a personalization strategy perspective the knowledge networks are limited in the sense that they do not support the knowledge processes of expanding and connecting. When the processes view guides the organizational design of the KN mechanism, it will naturally suppress the personalization strategy because this perspective is based on conflicting beliefs about how knowledge flows in organizations.

Advantages and limitations of knowledge networks as a mechanism for sustaining Software Process Improvement

9

References 1. Alavi, M. and Leidner, D. E.: Review: Knowledge Management and Knowledge Management Systems: Conceptual Foundations and Research Issues. MIS Quarterly, 25 (2001), 107-136 2. Benner, M. J. and Tushman, M. L.: Exploitation, Exploration, And Process Management: The Productivity Dilemma Revised. Academy of Management Review, 28 (2003), 238-256 3. Chrissis, M. B., Konrad, M. and Shrum, S.: CMMI: Guidelines for Process Integration and Product Improvement. Addison Wesley, (2003) 4. CMMI-Product-Team: Capability Maturity Model Integration (CMMI), version 1.1. CMMI for Software Engineering. Staged Representation (2002) 5. Davenport, T. H. and Prusak, L.: Working Knowledge - How Organizations manage what they know. Havard Business School Press, Boston (1998) 6. Germonprez, M. and Mathiassen, L.: The Role of Conventional Research Methods in Information Systems Action Research in Information Systems Research: Relevant Theory And Informed Practice, D. T. I. Bonnie Kaplan, David Wastell, A. Trevor Wood-Harperand Janice I. DeGross. Kluwer Academic Pub (2004) 7. Grant, R. M.: Prospering in Dynamically-Competitive Environments: Organizational Capability as Knowledge Integration. Organization Science, 7 (1996), 375-387 8. Hansen, M. T.: The Search-Transfer Problem: The Role of Weak Ties in Sharing Knowledge across Organization Subunits. Administrative Science Quarterly, 44 (1999), 82111 9. Hansen, M. T., Nohria, N. and Tierney, T.: What's Your Strategy for Managing Knowledge? Harvard Business Review, 77 (1999), 106-116 10. Hildreth, P. M. and Kimble, C.: The duality of Knowledge. Information Research, 8 (2002), 11. Lave, J. and Wenger, E.: Situated Learning: Legitimate peripheral participation. Cambridge University Press, Cambridge (1991) 12. Levinthal, D. A. and March, J. G.: The Myopia of Learning. Strategic Management Journal, 14 (1993), 95-112 13. March, J.: Exploration and exploitation in organizational learning. Organization Science, 2 (1991), 71-87 14. Mathiassen, L. and Pourkomeylian, P.: Managing knowledge in a software organization. Journal of Knowledge Management, 7 (2003), 63-80 15. Microsoft: Microsoft Solution Framework v3.0 Overview (2003) 16. Miles, M. B. and Huberman, A. M.: Qualitative Data Analysis. SAGE Publications, London (1994) 17. Nonaka, I.: A Dynamical Theory of Organizational Knowledge Creation. Organization Science, 5 (1994), 18. Polanyi, M.: The Tacit Dimension. Anchor Books, Garden City (1967) 19. Spender, J.-C.: Organizational knowledge, learning and memory: three concepts in search of a theory. Journal of Organizational Change Management, 9 (1996), 63-78 20. Swan, J., Newell, S., Scarborough, H. and Hislop, D.: Knowledge Management and Innovation: networks and networking. Journal of Knowledge Management, 3 (1999), 262275 21. Tiwana, A.: Affinity to Infinity in Peer-to-Peer Knowledge Platforms. Communication of the ACM, 46 (2003), 77-80 22. Wenger, E.: Communities of Practice - Learning, Meaning, and Identity. Cambridge University Press, Camebridge (1998) 23. Yin, R. K.: Case study research - design and methods. 3rd. Sage Publication, Thousand Oaks, CA (2003)