In today's organizations the common unit of work is the project. Turner defines a project as "an endeavor in which human, material, and financial resources are organized in a novel way, to undertake a unique scope of work, for a given specification, within constraints of cost and time, so as to achieve beneficial changes defined by quantitative and qualitative objectives" . Projects have moved from being simple phenomena to manage to more complex entities spanning geographical locations, multiple occurrences, and different organizational affiliations, with IT being the key enabler for the transformation. For instance, a co-located program involves multiple projects running at one location, whereas a distributed project is a single endeavor conducted from multiple locations. Finally, the most complicated scenario is multiple projects conducted at multiple locations. Complexities can be attributed to managing multiple interdependencies across time, space, and projects .
In the realm of project management, much of the effort in incorporating technology has involved fostering ubiquitous communication among members while facilitating knowledge exchange. Knowledge generated by projects can be categorized as knowledge in projects, knowledge about projects, and knowledge from projects . Knowledge in projects calls for a close look at insights generated within each individual project, such as schedules, milestones, meeting minutes, and training manuals. Individual project members need to know when, what, how, where, and why something is being done and by whom, with the goal being to promote efficient and effective coordination of activities. From the macro perspective, an organization must have an inventory of all projects under way at any given time, or knowledge about projects. This aids in the planning and controlling of resources to maximize utilities. Knowledge includes employee assignments to projects, return on investment, cost and benefit analysis, deadlines, and customer commitments and expectations. It is common for such knowledge to be generated at regular intervals, such as in weekly, monthly, or bi-monthly reports. Knowledge from projects is a post hoc analysis and audit of key insights generated from carrying out projects. This knowledge is a key determinant of future project success, as it aids organizational learning. These three categories call for distinct roles by IT to enable effective and efficient knowledge management.
Hansen and colleagues divide knowledge management approaches into two categories: codification and personalization . In codification, individual knowledge is amalgamated, put in a cohesive context, and made centrally available to members of the organization via databases and data warehouses. Here, we use a document-to-person approach on the premise that knowledge can be effectively extracted and codified. Personalization is the exact opposite; it recognizes the tacit dimension of knowledge and assumes that knowledge is shared mainly through direct person-to-person contacts. The role of IT is to facilitate communication among members through such tools as email and group support systems.
Figure 1. Centralized approach.
A close examination of the codification and personalization approaches led us to draw parallels to two popular models of computing: client-server and peer-to-peer (P2P). The client-server paradigm, wherein a centrally located resource is used by multiple clients to request services for task accomplishment, is common in most distributed computing environments. P2P is a rather recent computing paradigm in which all nodes can take the role of either client or server. A node can request information from any other node, or peer, on the network and also serve content. Due to the centralization of the main resource provider, client-server computing is similar to the codification strategy, whereas the distributed nature of P2P, in which each node owns and makes its resources available to the network, can be viewed as parallel to the personalization strategy. Hence, we term codification and personalization as the centralized and the P2P approaches, respectively, to knowledge management.
Figure 2. Peer-to-peer approach.
Here, we look at the implications of these approaches on the aggregation, transfer, and sense-making of knowledge in non-collocated work environments. Drawing on the strengths and limitations of each technique, we propose a hybrid model. We begin by comparing the approaches, focusing on three dimensions: sharing, control of, and structuring of knowledge.
Sharing. Many studies report that members of organizations fear that sharing their knowledge with the community at large makes them less valuable to the organization. As such, the idea of contributing to a central repository does not jibe well . In the centralized approach, there are inherent delays between the moment the knowledge is created in the minds of individuals and when it is posted to the repository. Individuals may delay posting not only for gatekeeping purposes but also to allow for confirmation of events, sometimes to the point of irrelevance. This defeats the concept of real-time availability of knowledge, as insights not captured immediately are lost. As individuals are more likely to store the draft notes and working documents of insights on their local repositories than on the main server, this concern is minimized in the P2P approach. Moreover, the P2P approach fosters dialogue among the various agents of the team and develops a spirit of community, as each agent interacts with peers to gain knowledge. Hence socialization and externalization is mandated, which is pivotal for tacit knowledge exchange .
Control. The centralized approach detaches the contributor from his or her knowledge. Once posted centrally, the author loses control over knowledge access and usage. In the P2P approach, each member of the organization retains his or her knowledge, as well as explicit control over its visibility. Members are connected to their peers in the organization and can choose what knowledge to share. Since individuals have control over their own knowledge repositories, they are less likely to view sharing of knowledge as a threat to their value.
Structuring. Knowledge contained in the central repository is structured on such dimensions as teams, products, and divisions, enabling faster access times to required elements. This facilitates the use of filtering and categorizing mechanisms for sifting. However, the nature of centrality calls for global filtering and categorizing schemas, which are not optimal in all cases. Setting global thresholds for relevance, accuracy, and other attributes for knowledge may lead to loss of knowledge, as insights considered important for one project may be lost due to filters. The significant costs in categorizing information by appending appropriate key words and metadata to knowledge prior to posting it are borne by everyone, whereas the benefits of better retrieval times are selectively reaped only by the users of knowledge. This asymmetry creates a particularly perverse motivation conundrum: those with potentially the least to gain (knowledge providers) pay the most for it, decreasing the attractiveness of the whole schema.
On the other hand, in the P2P approach, each agent may choose to invoke his or her unique coding and categorization scheme for inputting knowledge. While this allows for flexibility, it makes availability of shared context impossible. Such a system will become ill-structured over time, resulting in cumbersome search and seek times and irrelevant knowledge search results. The ease of effort through which knowledge can be made available to the network can act as a double-edged sword. While it ensures more real-time capture and dissemination of knowledge, it also leads to quality and validation issues. Moreover, when individuals add to the knowledge they download, multiple and inconsistent versions can result.
Figure 3. Hybrid approach.
In the centralized approach, members of the organization know where the knowledge objects reside and have the prerequisite tools and knowledge to access them, thus making for ease of use. These characteristics also make the centralized approach useful for storing structured knowledge about and from projects. Requirements for knowledge about projects do not change frequently. As such, having structured approaches for retrieval is facilitated via a centralized approach. But only a small percentage of the organization uses knowledge about projects for budget preparation, staff allocation, and other control purposes. Hence, storing such knowledge in a central repository is of minimal value to the remaining employees, or the majority of the organization. However, the P2P approach is not advisable here due to difficulties in filtering, categorization, and coordination of disparate knowledge sources.
Using the centralized approach to store knowledge from projects helps make lessons learned from past endeavors available to organizational members at large, but centralization is no panacea. Once again it may be difficult to contribute knowledge if appropriate categories do not exist or the repository is not amenable to customization. As the repository is built on the premise that all members of the organization need to access the knowledge base, care is taken to provide a shared context. This entails knowledge contributors making extra effort to ensure their thoughts and insights can be understood by their peers once entered into the knowledge repository. Developing a ranking mechanism to indicate the relevance of results becomes easier owing to the structured categorization of knowledge and the shared context. Thus, transfer of knowledge from the provider to the consumer is improved by the centralized approach.
The U.S. Army is a prime example of an organization taking a centralized approach to knowledge management. The Army Publishing Agency serves as one of the largest publishing houses in the U.S., producing thousands of publications. Today, the Army Knowledge Online (AKO) portal site contains most of these knowledge objects in digital format, giving Army personnel worldwide access to a unified collection of Army knowledge. AKO provides a single entry point into a growing knowledge management system that enables greater sharing among Army communities and enhanced communication between Army personnel and their civilian friends. AKO serves as a virtual home away from home for soldiers stationed around the world. With bases and personnel located worldwide, keeping in touch can be difficult. For security purposes, the Army relies on its own telecommunications systems rather than the local systems running in host nations like Germany, Italy, and the U.K. Most knowledge from and about past endeavors can be found on the AKO and is available to servicemen and women based on their degree of security clearance. The AKO has been highly successful in training new recruits and enabling faster access to lessons learned and protocols.
The solution in the U.S. Army example is unlikely to work in all situations. The recommended option for knowledge in projects is to use the P2P system. Adopting the centralized approach is inefficient, as much of the knowledge in projects is local to one site or to one project and as such, storing such information on a central repository is meaningless and irrelevant to nonmembers of the project. Since knowledge in projects is updated often, in many cases daily, to record details, such as project schedules, milestones, minutes of meetings, and training manuals, will result in overfilling the knowledge repository, high network traffic, and irrelevant search results.
John Deere, a global producer of tractors and other equipment, uses the personalization approach to knowledge management. Deere has established communities of practice (CoPs) for facilitating knowledge exchange. It has recognized hundreds of CoPs, approximately 65 of which use JD MindShare as a technology solution. Knowledge is exchanged within these CoPs via videoconferencing, email, and discussion groups. Using the personalization approach, Deere has experienced the challenges discussed in this article, including the difficulty of sharing knowledge among CoPs due to lack of shared context and varying schemas of knowledge representation.
A Hybrid Approach
To gain the benefits of the centralized and P2P approaches and overcome their limitations, we propose a hybrid model with two components. At the core of the system is the first component: a central repository holding popular knowledge (knowledge about and from projects). This repository serves as an index to the second component: knowledge available by peers (knowledge in projects). By storing knowledge about and from projects in a central repository, we ensure the following:
- Maintenance of a shared context, thus improving means of exploration of knowledge;
- Ease of access, as knowledge about projects is well structured and stored in a central repository;
- Ease of transfer of knowledge from projects throughout the organization;
- Enhanced validity of knowledge from projects, since only validated knowledge makes its way to central storage. Members whose knowledge is stored centrally can also be rewarded for its high value; and
- Easy identification of the source of knowledge about and from projects.
A centrally located index for knowledge in projects, the second component of the central repository, helps foster an efficient coordination mechanism, as this index contains the sources of knowledge in projects. It can serve as the knowledge dictionary to integrate individual knowledge. Common terms and categories can be assigned, along with facilities to serve as an organizational thesaurus.
We assert that knowledge in projects must be exchanged via P2P approaches. Knowledge capture becomes efficient and effective, as each project team can set up its own protocols, build categories, and develop filtering mechanisms. As each project is unique and each team is different, allowing for the establishment of flexible knowledge creation and exchange protocols is pivotal. Such an approach prevents the loss of knowledge considered relevant to one project, site, or stakeholder, since local parameters will not affect the global knowledge of the organization. Moreover, it promotes efficient sharing of knowledge among team members, as the fear of making one's insights available throughout the organization is absent. If an employee is interested in a knowledge object dealing with an ongoing project, he or she can refer to the index and obtain the source. The employee can then request that knowledge directly from the source.
Furthermore, maintenance of local repositories becomes simple, as each team is given the right to purge their local repositories periodically, therefore deciding which knowledge is relevant and useful and which is not. This not only prevents irrelevant and outdated search results, but also helps improve access times. It also helps circumvent the version control problem by constantly updating the repository while purging old and irrelevant knowledge.
Motorola, a major provider of wireless communications, semiconductors, and advanced electronic components, systems, and services, takes an approach to knowledge management that conforms to the hybrid model. Motorola's major business segments/sectors are the personal communication segment (PCS), network systems segment (NSS), commercial government and industrial systems segment (CGISS), semiconductor products segment (SPS), and other product segment (OPS). Motorola has an internal portal where knowledge objects (from and about projects) are stored in a central location for employee access and use across sectors. Examples of these documents include white papers, feature requirements documents, project test reports, and data reports. These documents are posted for the software/hardware features for a given product in a sector. Each employee can customize a portion of the portal, thus fostering the P2P approach. Functionality exists for storing their documents on the central repository with password-protected locations for confidentiality purposes. Passwords are issued by the owner (knowledge provider and document author) of the site, so if an individual first accesses the site he or she will be instructed to "contact the site owner/administrator for password." This guarantees that the material is not openly available, and people trying to access it have a conversation with the original source or owner of the material knowledge, just like a P2P exchange. Search functionalities allow employees to locate knowledge sources; however, a full-fledged indexing system is not present. Thus, Motorola follows a hybrid approach to knowledge management much like the one proposed here.
We have addressed the issues of knowledge management systems in non-collocated environments. We have specifically analyzed two common approaches to knowledge management and made a case for a hybrid model. Our insights have implications for practitioners, as one can consciously choose the right scheme for managing knowledge in projects, about projects, and from projects. For researchers, we have laid the foundation for inquiry into some key issues in distributed knowledge management. Questions of interest include empirically testing the theoretical framework using the different systems approaches. The role of context can be examined both from the global and the local perspectives. Our unique contribution is the analysis of how various system architectures affect knowledge exchange in, about, and from projects.
©2004 ACM 0002-0782/04/0400 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2004 ACM, Inc.