Bonifacio JUCS 2002

Summary

The authors take issue with the typical implementation of knowledge management projects from an organizational, social and technological point of view. They claim that the goal of most such projects is to create "large, homogeneous knowledge repositories in which corporate knowledge is made explicit, collected, represented and organized, according to a single -- supposedly shared -- conceptual schema" (p. 652). Such projects are typically patterned like so:

  • they require a Chief Knowledge Officer to manage knowledge flow across organizational units in such a way that all provide knowledge for the system, and all can benefit from the knowledge that is captured,

  • they utilize an organization-wide intranet to ensure universal accessibility within the organization,

  • they are built on a common set of definitions (language) and knowledge maps so that all knowledge to be entered into in the system can be represented in the same way,

  • the establishment of official communities of practice,

  • the design of processes for converting tacit knowledge to explicit knowledge, and

  • the construction of an enterprise knowledge portal that all use to access the captured explicit knowledge (p. 653).

There are fundamental flaws in the assumptions underlying this pattern. The majority of knowledge is tacit: it is contextual, social and subjective. Making tacit knowledge explicit is to make it more abstract and general, and to remove it from its context. While this "assumption is coherent with the traditional organizational models and paradigms of control" (p. 653), it assumes that knowledge can, in fact, be represented objectively and still retain value. Different individuals and groups have different ways of interpreting the world; they have different interpretation schema. Different schemas can produce different interpretations. "In general, interpretation schemas are only partially reducible to each others [sic]" (p. 655) -- the value of knowledge may be lost or reduced when translating it to another interpretation schema. Since the typical pattern of KMS implementation (as described) involves the forcing of knowledge into a privileged schema -- that defined by the CKO -- users of the resulting organization-wide KMS are certain to be dissatisfied: the knowledge in the system will either be perceived to be irrelevant (because the privileged schema is alien to the user and not deeply understood) or the KMS will seem oppressive (because the privileged schema is understood by the user but is not acceptable). For these reasons, many KMS projects implemented in the typical pattern suffer from abandonment by their targeted users (p. 654).

The authors propose a different model for KMS implementation: distributed knowledge management, based around knowledge nodes. Since each organizational subgroup (typically functional groups or project teams) is likely to have its own interpretation schema, let them manage their own local knowledge in a way that seems best to them. These subgroups are the knowledge nodes. They link those separate knowledge management systems into a organization-wide system via the use of software agents which can translate knowledge between different schemas. This translation is accomplished via conceptual maps that map the schema of foreign knowledge nodes into a node's own context. When a search for information is undertaken by a user in one node, that node's agent presents its request to the other agents using the node's own schema. The other agents then translate the other node's schema into their own node's schema to search their local knowledge stores.

The authors illustrate the viability of their implementation model with a case study regarding an Italian bank. They use the study to give evidence for the natural existence of knowledge nodes, and create a hypothetical set of schema and agents to create the distributed knowledge network.

Issues

This article is relevant to Dalkir's chapter on "The KM Team" [2] because she proposes a personnel heavy, top down approach to organization-wide knowledge management; this type of management is precisely what this article argues against.

Bases of distributed knowledge management

The authors base their distributed knowledge management implementation on two general principles: the principle of autonomy, and the principle of coordination.

  • The principle of autonomy says that each subgroup of an organization should be given a high degree of autonomy in the management of its own local knowledge, letting them devise the most appropriate interpretation schema to their context (p. 656).

  • The principle of coordination says that each such subgroup must be able to exchange knowledge without the use of a common schema, but instead by translating the schema of other subgroups onto its own (p. 656).

Social construction of reality

The authors talk at length of interpretation schemas and the importance of context, as well as the social construction of knowledge: there are communities of knowing which are defined by a certain agreement among its members on how the world and events in it are to be interpreted coherently. This is a restatement of the idea of social constructionism, brought to prominence by Berger and Luckmann in their 1966 book The Social Construction of Reality. From Wikipedia's entry on social constructionism:

  • The focus of social constructionism is to uncover the ways in which individuals and groups participate in the creation of their perceived reality. It involves looking at the ways social phenomena are created, institutionalized, and made into tradition by humans. Socially constructed reality is seen as an ongoing, dynamic process; reality is re-produced by people acting on their interpretations and their knowledge of it. [1]

If one takes as given that organizations are made up of separate subgroups, then the idea of instituting a universal way of interpreting the world throughout the organization can be seen as sub-optimal. This is because each subgroup constructs their own version of reality which is presumably efficiently tailored to the needs and goals of the subgroup and which may be different than that of other subgroups.

Critique

The authors are trading local efficiencies (knowledge nodes use their own knowledge most effectively if they retain their own perspective and manage their own knowledge accordingly) for global inefficiency and effort (someone has to write the agents, and the concept map translations). They do some handwaving on the critical concept map translation part and say the agents should be able to arbitrate between contexts (p. 660) without describing how this is to be done other than saying that the knowledge nodes should take part in definition. This is at least an O(n) problem for each group (disregarding inter-group communication problems, and the translation problem in the first place) because each group must be able to translate every other group's interpretation schema into their own. For the organization, the effort is O(n<sup>2</sup>) although the effort is distributed across the groups. I fear that this effort is exactly the type that will not be expended, as it is subject to the prisoners dilemma -- everyone benefits only if everyone participates. Each group cannot be assured that the other groups will reciprocate if they create translation maps for each of the other groups in the organization, and it may be seen to be in their interest simply to not expend that effort. And since local interpretation schema are dynamic, someone needs to be responsible for updating the translation maps.

References

[1] http://en.wikipedia.org/wiki/Social_constructionism
[2] Dalkir, Kimiz, Knowledge Management in Theory and Practice, Elvsevier Butterworth-Heinemann: Amsterdam, 2006.