Until recently what was missing was an acknowledgment that not all knowledge can be captured, codified, and stored. This has now been recognised and has started to receive attention. Interestingly, it has also been acknowledged within the community that could be seen to be most firmly entrenched in the capture-codify approach (Buckingham Shum, 1998), where a recognition has been made of the importance of socially constructed knowledge:

There appeared to be a strong sense at a recent symposium on AI's role in KMthat formal representation of knowledge seems to have a limited role to play in organizational knowledge management, with the emphasis shifting to supporting the social, coordinated processes through which knowledge is constructed (p. 74).

This recognition that not all knowledge can be treated as an object has raised other difficulties. In the first place, if there is knowledge that cannot be objectified, this has implications for the management of knowledge. The view of KM as quantifying, measuring, and controlling is no longer sufficient and the definition of KM needs to be broadened. This has led to a further difficulty. There is now a wide variety of KM definitions to the point where there is no real consensus as to what KM is. This was demonstrated by Alavi and Leidner (1997), who investigated KM views among Chief Information Officers (CIOs), Information Systems (IS) Managers, and functional area executives, and found it to be the case even among business practitioners. In fact, they found three strands:

  • An information-based view that is often the dominant view, that is, of knowledge being codifiable, then being captured, stored, and transmitted.

  • A technology-based view that sees KM in terms of IT solutions.

  • A culture-based view that is based on a different view of knowledge.

Alavi and Leidner's (1997) survey reveals the tension between the views. The information-based view is the traditional view of KM, treating knowledge as an object. The technology-based view is a continuation of this but also reflects the view of those people who consider that KM consists of simply implementing technology in order to manipulate the abstracted knowledge. The shift to recognising the importance of knowledge that cannot be captured is reflected in the third view that places more importance on "learning." This still only reflects the beginning of a shift in views; the long list of examples in the information- and technology-based views (Table 1) demonstrates the longestablished work that has been done in this field. The four examples given for the culture-based view show that, while people finally recognise the importance of this area, there is still some way to go to understand fully what is needed to manage this type of knowledge.

Table 1: Perspectives on the Meaning of Knowledge Management (Adapted from Alavi & Leidner, 1997, p. 13)




Actionable Information
Categorizing of Data
Corporate Yellow Pages
Filtered Information
Free Text and Concepts
People Information Archive
Readily Accessible Information

Data Mining
Data Warehouses
Executive Information Systems
Expert Systems
Intelligent Agents
Search Engines
Smart Systems

Collective Learning
Continuous Learning
Intellectual Property Cultivation
Learning Organisation

Karl Sveiby (n.d.) explains the difference in more detail. He identifies two distinct approaches to KM, which correspond with the information-based and culture-based views encountered by Alavi and Leidner (1997):

  • Track 1: This track views knowledge as information. To researchers and practitioners in this field, it is an object that can be captured and handled in an information system. People in this field will have a background in computer science or information science and will be involved in AI and information management systems. Sveiby states that this is a new and fast growing track, based on developments in IT.

  • Track 2: Sveiby feels that this is an old track that is not growing very fast. People in this field regard knowledge as processes, constantly changing skills, and expertise, and have their backgrounds in philosophy, psychology, sociology, and management. This track corresponds more closely with the culture-based approach.

Sveiby also categorises KM on two levels:

  • the individual level where the focus in research and practice is on the individual; and

  • the organisational level where the focus is on the organisation.

This view is summarised in Table 2.

Table 2: Sveiby's Knowledge Management Perspectives

Knowledge Management



Knowledge = Object

Knowledge = Process

Organisation Level


"Organisation Theorists"

Individual Level

"AI Specialists"


The positive aspect of this view is that it recognises the different strands, but the situation has since changed. The information track was the fastest growing as a result of developments in IT; however, although the knowledge-as-process track subscribes to an older view of knowledge, it is only recently that its importance to the organisation has been recognised. Sveiby's identification of the strands is helpful but still does not explain the different challenges that this poses to KM—this type of knowledge is so tied to people that when they leave an organisation, the knowledge leaves with them, despite the best efforts of knowledge engineers with their rules and formulas.

This poses the question that if this knowledge cannot be captured and codified, what does "management" mean? There is, therefore, a need for a more all-encompassing approach to defining KM, moving from quantification, measurement, and control to involve a generative aspect (for example, learning and knowledge creation), sharing and application, retention, trying to organise it, and ensuring it is available where and when it is needed. The American Productivity and Quality Center (APQC) with the consulting company Arthur Andersen (now known as Accenture) has jointly developed a KM framework that shows the importance of managing all knowledge resources, and not just concentrating on the capture/codify/store cycle.

This diagram represents all the elements that APQC and Arthur Andersen believe require consideration when managing knowledge. It acknowledges the views of knowledge as object and process. It views KM as dynamic processes, and therefore shows two "orbits," one that shows the organisational enablers of technology, culture, leadership, and measurement. The second, inner orbit shows KM processes. Identify, collect, and organise refer more to the "knowledge as object," or "knowledge as information" aspect and relate to the capture-codify-store approach that has thus far been dominant. Create, apply, and adapt are KM processes that are applied when knowledge is viewed as a process. Sharing knowledge is necessary for all knowledge. How the knowledge is shared differs according to the type of knowledge.

A positive aspect of the diagram is the recognition of enablers, including technology. This is in contrast to some of Alavi and Leidner's (1997) respondents who regarded the implementation of IT as being KM itself. Figure 1 suggests that IT is simply an enabler, not a solution, and that more than IT is essential. It shows that culture is important—for example, if an organisation has not or cannot create a sharing culture, then KM initiatives will not be successful. In order to create such a culture, effective leadership is important. Measurement as an enabler is perhaps the weakest point of the diagram. Many people have the view that if something is to be managed, it must be able to be measured. Applying measurement criteria to KM initiatives may convince some skeptics of their use; however, quantification is not easy with a subject such as knowledge and as such could perhaps only be applied to tangibles such as training courses, books, databases, and database transactions—all of which are instances of "traditional" KM. Measurement would be difficult to apply to the management of this other type of knowledge.

click to expand
Figure 1: Knowledge Management Framework (From Finerty, 1997; O'Dell & Jackson Grayson, 1998)

The wider definition of KM helps—it shows the need for more than quantification and control by demonstrating the need for the generation of new knowledge, for sharing, for distributing, and for retention—but it still does not explain how this might be done for knowledge that cannot be captured and stored. This poses further questions for the KM community—what knowledge can be stored and what is the knowledge that cannot be captured?

These questions have led to a confused picture of knowledge within the KM community as researchers and practitioners have sought to define the knowledge that can be captured and that which cannot.

KM Views of Knowledge

In exploring more subtle types of knowledge the prevalent view in KM has been that such knowledge is "tacit" (Nonaka, 1991; Polanyi, 1967) and that codifiable knowledge is explicit.

The interest in tacit knowledge in KM has come about because of an Eastern influence in the field of KM. Cohen (1998) points to Nonaka and Konno (1998) as an example of the differing approaches between East and West.

Table 3 shows how the Western view of KM reflects the dominant view of knowledge as being an object that can be captured and codified. The Japanese or Eastern view places more of an emphasis on tacit knowledge.

Table 3: West-East KM Contrasts (Adapted from Cohen, 1998, p. 24)



Focus on Explicit Knowledge

Focus on Tacit Knowledge



Knowledge Projects

Knowledge Cultures

Knowledge Markets

Knowledge Communities

Management and Measurement

Nurturing and Love

Near-Term Gains

Long-Term Advantage

Explicit knowledge is knowledge that is easily expressed, captured, codified, stored, and reused. It is easily transmitted as data and is therefore found in databases, books, manuals, reports, and messages. Tacit knowledge, however, according to Nonaka (1991) is something very personal and difficult to articulate (and therefore difficult to communicate to other people). Nonaka also explains that it is rooted in action and consists "partly of skills—the kind of informal, hard-to-pin-down skills captured in the term ‘know-how’" (p. 98).

He described how tacit knowledge is shared in a "spiral" of knowledge (Figure 2).

click to expand
Figure 2: Nonaka's Spiral of Knowledge

He explained this with the example of a developer who was part of a team trying to design a bread-making machine. As the team had problems, one of the developers had the idea of apprenticing herself to a master baker. She trained with the master baker, observing the special way he stretched the dough. She returned to her team, and after a long period of trial-and-error, the team came up with the successful specifications for the design of the machine. Nonaka described the developer's period of "apprenticeship" with the baker as the time when she was being "socialised" into the craft; that is, there was tacit knowledge transfer. Returning to her team, he says she was able to "articulate" or externalise the knowledge she had acquired, moving it from her tacit knowledge to explicit knowledge in order to communicate it. The following stage in the spiral is where tacit knowledge of one field is combined with tacit knowledge of another field, as in the example of the externalised tacit breadmaking knowledge being combined with the knowledge of the development field. Finally, the new knowledge is internalised and becomes tacit knowledge. Nonaka emphasises that the sharing of tacit knowledge is exchanged through joint activities and requires physical proximity, and that externalisation needs tacit knowledge to be translated into forms that can be understood by others. However, an important aspect of tacit knowledge is that it is, by definition, extremely difficult to articulate. In fact, there are different opinions about whether tacit knowledge is articulable at all. Some commentators are of the opinion that tacit knowledge can, in fact, be captured (Huang, 1997); some feel it is merely "difficult" to articulate (Teece, 1998); others feel it cannot be codified without being invalidated (Buckingham Shum, 1998); whereas others feel it simply cannot be captured or codified at all (Chao, 1997; Leonard & Sensiper, 1998; Star, 1995). For example, Goguen (1997) gives the following examples of tacit knowledge being inarticulable:

People may know how to do something, without being able to articulate how they do it. In the social sciences, this is called the say-do problem. Some examples are riding bicycles, tying shoe laces, speaking languages, negotiating contracts, reconciling personal differences, evaluating employees, and using a word processor (p. 33).

In fact, Nonaka's (1991) examples are impossible to articulate—a master craftsman with rich experience who cannot articulate the principles behind what he knows or does; the master baker who can teach someone the principles only by taking them through a period of apprenticeship. In Nonaka's spiral of knowledge, tacit knowledge is transferred by the interpersonal interaction of the apprenticeship element. The tacit knowledge is not articulated expressly; rather, the learner develops the tacit knowledge in a situated environment under the guidance of a mentor by becoming immersed in the practice itself. The weakness of the spiral comes in the tacit-explicit stage. If tacit knowledge is inarticulable, this stage simply cannot work.

Nonaka's (1991) suggestion for overcoming this problem is the use of metaphors, which is similar to the use of stories. Orr (1990), in his study of photocopier repair engineers, described how an engineer encountering a problem not covered by the procedures found in the manual would consult with a supervisor or colleague. They would discuss the problem, try new approaches based on experiences of similar problems, and usually manage to solve the problem. The story of the problem and its solution would gradually circulate among the community of copier repairers and would become part of the stock of knowledge. They referred to this as the telling of "war stories." Goldstein (1993) encountered something similar when observing supermarket managers using computer-based data. He noticed they would examine the data, look for unexpected results, and try to make sense of them. They would then share this with peers by telling "stories." This still leaves a problem—the metaphor and the story are being "articulated," as in explicit knowledge. This raises the questions of what knowledge is being shared in the story, how it is being received by the learner, and what form does the knowledge take in the story?

Perhaps because Nonaka's (1991) tacit-explicit distinction does not answer the questions, other KM researchers and practitioners have made other distinctions. Rulke, Zaheer, and Anderson (1998) focused on the knowledge of an organisation that they termed transactive knowledge, which was made up of the organisation's self-knowledge (knowing what you know) and resource knowledge (knowing who knows what).

Conklin (1996) uses the terms "formal" and "informal" knowledge to distinguish between the types of knowledge of interest to KM practitioners. He describes formal knowledge as that found in books, manuals, and documents, and which can be shared by training courses. He observes that it is easily and routinely captured in organisations. Informal knowledge, on the other hand, is the knowledge that is applied in the process of creating formal knowledge. As examples, he gives ideas, assumptions, decisions, and stories. It is processoriented and is difficult to capture. Part of this is symptomatic of another distinction that is made by the KM community; that is the distinction between knowledge as information and knowledge as process. In this view, knowledge is regarded by some as being a superior kind of information and by some as being the process of knowing.

Kogut and Zander (1992) differentiate between information and knowhow, and Seely Brown and Duguid (1998) make a similar distinction with "know-what" and "know-how." They explain that "know-what" is explicit knowledge and easily shared, whereas the "core-competency" aspect of organisational knowledge needs more. It needs the "know-how" in order to put "know-what" into practice. However, even in this distinction there is an inconsistency in that know-how can have an "explicit" component, for example, procedures can be seen as a codified form of "know-how" that guides people in how to perform a task.

All of these distinctions are viewed in the form of opposites, but none of them have proved to be truly satisfactory in that they still seem to be limited, and in some there are distinct inconsistencies. Leonard and Sensiper (1998) try to tackle this by moving from a dichotomy to a continuum. They define tacit knowledge as "semiconscious and unconscious knowledge held in people's heads and bodies" and place it at one end of the spectrum. They define explicit knowledge as "codified, structured, and accessible to people other than the individuals originating it" and place it at the other end, with most knowledge somewhere in between.

Even this view, however, despite terming it a continuum, still places some types of knowledge at the ends of the spectrum and views them as opposites.

This move to exploring the management of less structured knowledge is explained by von Krogh (1998) as a shift in perspectives. He describes the more recent emphasis as being from a "constructionist" perspective, as opposed to the cognitivist perspective. He explains that the constructionist perspective regards knowledge as an act of construction or creation, with tacit knowledge being highly personal and difficult to express. As examples, von Krogh offers "physical skills, such as putting the movements together in a highprecision luxury watch, as well as perception skills, such as interpreting a complex seismic readout of an oil reservoir" (p. 134).

Von Krogh's (1998) constructionist view is based to a large degree on earlier work he did comparing the representational or cognitivist view of knowledge with a particular type of constructionist view—autopoiesis. Autopoiesis is a view of knowledge that is based on observations of cell reproduction in neurobiology. According to von Krogh, Roos, and Slocum (1996), it does not view the world as a state that can be represented:

Rather, that cognition is a creative act of bringing forth a world. Knowledge is a component of the autopoietic (self-productive) process; it is history dependent, context sensitive, andat the individual level, knowledge is not abstract but rather is embodied in the individual (p. 163).

A key aspect of autopoiesis is that it is self-referential; i.e., it includes potential future knowledge as well as past knowledge. People use established knowledge to determine what they see and to make distinctions in what they see. They also use what they already know to select what they are looking for in the environment. In other words, we use what we already know to decide what to look for in our surroundings (von Krogh, Roos, & Slocum, 1996). Changing those surroundings is undertaken by making different and finer distinctions and new interpretations (Vicari, von Krogh, Roos, & Mahnke, 1996), thereby evolving to higher states of knowledge.

The differences between the traditional and autopoietic perspectives are summarised in Table 4.

Table 4: Autopoietic vs. Traditional View of Knowledge (Adapted from Vicari et al., 1996)

Autopoietic view

Traditional view

Knowledge is creational and based on distinction making in observation

Knowledge is representation of a pre-given reality

Knowledge is history dependent and context sensitive

Knowledge is universal and objective

Knowledge is not directly transferable

Knowledge is transferable

Underlying the arguments of von Krogh, Roos and Slocum (1996) regarding the use of autopoiesis theory is the idea that knowledge is socially constructed. This is perhaps the central debate between the views of knowledge in the field of KM, that is, knowledge as data and knowledge as reasoning and culture. In anthropological, socio-psychological, and sociological work, knowledge tends to be regarded as a social product, whereas cognitive psychology has tended to emphasise representational and perceptual knowledge (Aadne, von Krogh, & Roos, 1996; von Krogh & Roos, 1995).

Traditional cognitivism assumes human cognition is a form of information processing and views the world as "pre-given" with the aim of creating the most accurate representations of the world. It regards knowledge as being representations of the world held in the minds of individuals in the form of knowledge structures, or schema constructs (Walsh, 1995). In this form, it regards the individual as an information processor. Von Krogh and Roos (1996) observe that representationism has formed the basis of much strategic management literature and has contributed to the view of knowledge as information.

There is, however, a different view in cognitive psychology that recognises the importance of the social context and culture. Bruner (1990) argues that we should move away from the notion of the individual merely as a processor of information and move the emphasis to meaning and how this is negotiated in a community, as individuals cannot exist independent of culture. He observes that symbols and codes (for example, the digits we use, mnemonics) that humans use to create meaning exist in and are developed in culture and communities. Wenger (1998), too, stresses that "information stored in explicit ways" is only a small part of the picture and that knowing is primarily something that comes about by active participation in communities. Hutchins (1995), in developing his theory of Distributed Cognition, also noted that looking for knowledge structures inside the individual fails to recognise that human cognition always takes place in a complicated social cultural environment and must therefore be affected by it. He emphasises that we must see cognition as part of a cultural process.

Although von Krogh (1998) describes the shift in perspectives as being from a cognitivist/representational approach to a constructionist view, the opposite views described earlier generally seem to still be rooted in the cognitivist/representational approach. This might explain why the management of the unstructured knowledge is proving such a challenge—although it is so difficult (perhaps impossible) to articulate, capture, codify, and store, the same approach is being taken. Attempts to manage such knowledge seem to focus on making tacit knowledge explicit and converting less structured knowledge to a form whereby it can be captured. There is clearly a need for a different view of knowledge in order to overcome this challenge of managing knowledge that cannot be captured, codified, and stored.

Viewing Knowledge Differently

Cook and Seely Brown (1999) have also come to the conclusion that a new view of knowledge is required; that is, they emphasise the need for a move from possession to practice. As a result, they emphasise that tacit and explicit knowledge are distinct types of knowledge and it is impossible to move from one to the other; that is, tacit knowledge cannot be made explicit. This is unlike the general KM approach that is to try to make tacit knowledge explicit in order that it may be more easily used and managed.

Cook and Seely Brown (1999) retain the notions of tacit and explicit but note that each does work that the other cannot. To illustrate this point, they use the example of riding a bicycle. In order to ride a bicycle, one must have (tacit) knowledge of how to remain upright. They emphasise that this is not the activity of riding, but knowledge used in riding. The rider still has the tacit knowledge even when (s)he is not riding. It is the tacit knowledge that enables the rider to remain upright, and this is something that explicit knowledge cannot do. Instructions could be given regarding how to ride a bicycle. This would be explicit knowledge that could be given to a novice, but it would not be able to do all the work that would enable someone to know how to ride. The tacit knowledge can only be gained by actually riding the cycle. Although each form of knowledge does work that the other cannot do, each can be an aid in acquiring the other; for example, the explicit knowledge may help a beginner get started while (s)he acquires tacit knowledge. It is important to note, however, that there is no conversion—one can be used to generate the other.

Cook and Seely Brown (1999) also add in another dimension—that of "knowing." In their bicycle example, they note that tacit and explicit knowledge are not sufficient to develop the ability to ride a bicycle. The act of riding itself is also necessary. They call that which is possessed "knowledge," emphasising that it is used in action but is not itself action. Knowing, on the other hand, is described as being part of action. They refer to it as doing "the epistemic work that is done as part of action or practice" (p. 387). Knowledge and knowing are not opposites—they are complementary to each other. Cook and Seely Brown refer to this as being "mutually enabling."

Knowledge is a tool of knowing,knowing is an aspect of our interaction with the social and physical world, and that the interplay of knowledge and knowing can generate new knowledge and new ways of knowing (p. 381).

These authors consider that the interaction between knowing and knowledge is generative—it leads to new knowledge being created. Using this approach they view Nonaka's (1991) example of the bread machine from a different angle. In Nonaka's account, the designer obtained the tacit knowledge from the master baker and then articulated it (made it explicit) for her colleagues in the design team. Cook and Seely Brown (1999) reject the idea that the tacit knowledge was made explicit. Rather, they consider that in addition to the knowledge (tacit and explicit) of all of the members of the group, there was also knowing—the way the group members interacted with each other, the dough, and the machine parts, that finally led to the creation of new knowledge.

Cook and Seely Brown's (1999) work is interesting and useful for the move to find a different view of knowledge from that used by most KM practitioners and researchers. The extra aspect of "knowing" that has been added to the picture is helpful in that it reinforces the view that the tacit-explicit pairing is not sufficient to explain the knowledge that cannot be captured and shows that an epistemology of practice is also necessary. However, knowledge is also still regarded as something that is possessed, and the "practice" part is something extra. What is needed is a view of knowledge that incorporates the practice, the socially constructed, the elusive knowledge. Perhaps in the first place a more general distinction is needed as a working hypothesis. For the purposes of this book, the terms "soft knowledge" and "hard knowledge" will be used. As a starting point, hard knowledge will be used to describe that knowledge that can easily be captured, codified, and stored. Soft knowledge will be used for the elusive knowledge that is posing a challenge to KM. However, soft knowledge in particular needs more examination, for it has already been mentioned that the predominant view of this type of knowledge still has its roots in representationism. However, von Krogh (1998) has shown that there is a shift to constructionism.

We have seen in this chapter that there is a shift to looking at soft knowledge, but that the attempts to define it have resulted in a variety of opposites, all of which are hard or soft. The management of hard knowledge is well established, but the management of soft knowledge is posing a challenge to KM. In Chapter III, therefore, we will explore in more detail the notion of soft knowledge in order to refine the concept and to attempt to understand the relative roles of representationism and constructionism.