Developing the core principles of responsible knowledge management (rKM): Section 2.5 – Knowledge as an asset: problems & inconsistencies

This article is Section 2.5 of Chapter 2 of a series featuring my Master’s thesis The Emerging Concept of Responsible Knowledge Management (rKM): Identifying and Formulating the Core Principles of rKM.

Knowledge proves challenging to define incontestably. Despite being the foundational concern of knowledge management (KM), its boundaries remain porous, its meanings challenged, and its role shaped by its uses. This section explores the conceptual tensions that arise when KM attempts to label and manage knowledge, particularly when defined as an asset. It addresses the ambiguity between tacit and explicit knowledge, the assumptions rooted in the DIKW model, the preoccupation with measuring the performance of knowledge as an asset and the consequences of these orientations on the meaning derived from KM. These tensions are central to understanding KM’s intellectual complexity.

In the Western philosophical tradition, knowledge is usually defined as ‘justified true belief’¹, and as such, subject to empirical validation. Unfortunately, many concurrent actualities complicate this seemingly simple statement. First, what counts as justified true belief is open to question and depends on the perspective; second, knowledge may also be viewed as socially constructed; and third², “power influences what counts as – or is accepted as – real knowledge.” ‘Justification’ is often culturally and politically mediated, and what is considered ‘true’ may shift over time or between communities. In practice, knowledge resists containment. It moves through implicit understanding, informal practice, and discursive frames. It is always situated, partial, and, in some forms, ephemeral. This makes the notion of knowledge as neat unit of commodifiable fact more of a strategic aspiration than an ontological truth.

Nevertheless, defining knowledge as an asset has enabled KM to develop managerial tools and practices, but it has also generated persistent tensions: between tacit and explicit, between knowledge and information, and between lived practice and codified artefact. These tensions suggest that the asset metaphor is both enabling and limiting, raising questions about whether the most vital aspects of knowing can ever be contained within it.

2.5.1 The explicit/tacit binary

History proves that KM chooses to focus on the distinction between explicit and tacit knowledge, terms used to describe³, respectively, knowledge that can be codified and shared through formal means, and knowledge that is personal, experiential, and difficult to articulate. Following the concept’s operationalisation by Nonaka and Takeuchi⁴, this distinction is frequently invoked to justify KM practices aimed at integrating, i.e. capturing and storing knowledge, especially to legitimise proprietary claims over the more inaccessible forms of knowing.

However, as Roberts reminds us⁵, citing Polanyi⁶, “all knowledge is either tacit or rooted in tacit knowledge. A wholly explicit knowledge is unthinkable.” This suggests that even the most structured or formalised knowledge still relies on a degree of tacit understanding whether in interpretation, context, or application. The captured knowledge artefacts may represent parts of the whole, but they are not themselves knowledge without human engagement. This raises the question of whether the relentless effort to divide knowledge into explicit or tacit components is ultimately worthwhile, or whether it distracts us from more meaningful inquiries. A human is always part of the loop; the larger and more inclusive that loop becomes, the more dynamic, plural, and nuanced the knowledge it can support. Attempts to reduce knowledge to manageable fragments may offer the illusion of clarity, but at the expense of wisdom, risking the loss of the forest for the trees. Perhaps it would be more conducive to accept that knowing is an activity whose explicitness is beyond full capture and whose interpretation always needs frames of reference that exceed the facts.

In this vein, Wilson⁷, one of the most acerbic critics of KM, serves up some uncomfortable ‘truths’ about the concept of tacitness, and his understanding of the difference between the categories of knowledge and information. In no uncertain terms, Wilson claims that Nonaka and Takeuchi erroneously (on purpose or by accident) ‘confused’ tacit knowledge with implicit knowledge. Drawing on Polanyi’s original formulation Wilson argues that “‘tacit’ means ‘hidden’, …hidden even from the consciousness of the knower.” Tacit knowledge is, therefore, not within the range of capture or conversion through the SECI model. What the model might address is more accurately termed implicit knowledge, something so far unexpressed but still in the realm of expressible.

Indeed, if we follow Wilson’s critique to its logical end, tacit knowledge may best be conceptualised not as a convertible asset but as a set of pre-reflective heuristics akin to what Kahneman^8,9 describes as ‘System 1’ thinking: fast, automatic, and often opaque decision-making processes. These orienting dispositions are not usually externalised or codified; rather, they are shaped by biography, context, and embodied experience.

Furthermore, Wilson is quite adamant that once knowledge is expressed, it becomes information. A codified, extracted message can be assimilated and incorporated but it is not the knowledge itself; “the messages can never be exactly the same as the knowledge base from which the messages were uttered.” This challenges the very core semantics of KM. If knowledge cannot be extracted, stored, or transferred in its full form, then what exactly is being managed by KM processes?

KM is seemingly unencumbered by these connotative disputes, continuing to worship at the altar of models like SECI. This ‘vital’ knowledge creation instrument rests on an idealised spiral of knowledge conversion from tacit to explicit, from individual to collective, and back again. While it has been praised for introducing dynamism into KM discourse, it also presupposes a self-reinforcing system in which knowledge flows are continuous, manageable, and endlessly generative, so long as the organisational conditions are correctly designed.

But this optimism carries with it an implicit metaphor of the organisation as a machine for perpetual innovation, in which knowledge workers convert their knowing into artefacts, codifications, and actions in service of institutional goals. This logic resembles the rhetoric of neoliberalism, where all human activity is harnessed toward productivity, and the systemic frictions of emotion, ambiguity, dissent, or ethical deliberation are treated as inefficiencies to be designed out. This fantasy of seamless conversion pays no heed to the partial, relational, and often resistant nature of knowing, or to the ethical responsibility of attending to these limits.

The persistent pursuit of the tacit/explicit debate raises the question of whether it is the most relevant distinction to focus on. If knowledge is better understood as socially constructed and continually reinterpreted in dialogue, then this fixation is missing the point. Rather than a final recorded truth, knowledge may be closer to an ongoing resonance – always partial, always in motion.

2.5.2 The data information knowledge wisdom (DIKW) model

Some terminology is nevertheless required for the strategic handling of the various aspects of knowing. For this reason, KM has eagerly adopted the information science view of conceptualising knowledge as part of its own DIKW model, traditionally presented as a pyramid that assumes a progression. However, these four concepts might be more fruitfully understood as a kind of tetrarch: distinct domains with overlapping authorities, each governing a conceptual territory that is in constant negotiation with the others.

The DIKW hierarchy is widely used to conceptualise the relationships between the aforementioned four categories, despite its origins remaining obscure¹⁰ (Lambe traces it to Harlan Cleveland¹¹ who may have been inspired by T.S. Eliot¹². Commonly also attributed to Ackoff¹³)..

The linear and hierarchical representation is problematic, argues Roberts¹⁴, who prefers to see the relationship between data, information, knowledge, and wisdom as “multidimensional, recursive, and/or random.” Data, she explains, is a collection of observations, measurements, or facts, the raw material from which information is formed; information emerges when data are organised into meaningful patterns; and knowledge is something that when applied to information, enriches its understanding through experience, familiarity, and learning. Therefore, knowledge creation depends on information; likewise, the development of relevant information requires the application of knowledge. Referring to McKenna¹⁵, Roberts adds that wisdom lies in recognising the limits of knowledge.

When viewed from this angle, the knowledge asset indeed begins to resemble information. This is Dillon’s¹⁶ pragmatic counter to Wilson’s existential refusal to treat knowledge as something manageable. He labels Wilson’s ‘information’ as “recorded knowledge.” The tension between Wilson’s philosophical critique and Dillon’s pragmatic response reflects one of the central theory-practice dilemmas of KM discourse. Wilson is philosophically correct in asserting that knowledge, properly understood, cannot be managed: it is embodied, situated, and shaped by individual consciousness and social context. To treat it as an object is to engage in a category error.

Yet, Dillon is practically correct. If we are to engage with the organisational realities of learning, memory, and judgment, we must be able to talk about them, to name and structure that domain, however imperfectly. This requires some form of terminological packaging. We may not be able to manage knowledge per se, but we can manage artefacts, environments, and practices that support knowing. For Dillon, “recoded knowledge is inert” and KM “refers to processes that manage this inert stuff.”

He argues that information systems must handle different levels of complexity, and that labelling some content as ‘knowledge’ serves to highlight those instances where information becomes deeply contextualised, synthesised, and perhaps even anticipatory. Thus, while he does not provide a philosophically rich account of knowledge as embodied or tacit, he nonetheless defends the need for the term ‘knowledge’ to describe this highest level of abstraction in automated systems and institutional contexts.

It is clear from Dillon’s text, however, that he acknowledges a transformation occurs when recorded knowledge is taken up by a subject. By referring to recorded knowledge as ‘inert stuff’, Dillon highlights its passive, externalised nature. However, he also indicates that this inert material becomes something else in the process of being interpreted, assimilated, or used. He does not spell this out in phenomenological terms, but the implication is there: knowledge is not fully actualised until it enters into a cognitive or interpretive relationship with a knowing subject.

This affects his understanding of what knowledge is in a subtle but telling way. Dillon’s framework accepts that knowledge, once recorded, loses its immediacy and becomes artefactual, but at the same time, he implies that such artefacts have the potential to regenerate knowledge through engagement. In this view, knowledge exists in two states: as static artefact and as dynamic process. His defence of the term ‘knowledge’ for recorded content is thus pragmatic. He retains the label because it signals a higher level of information complexity, not because he believes the knowledge-as-lived-experience resides in the object itself.

So while Dillon’s primary concern is functional categorisation within information systems, he does not completely deny that knowledge properly resides in people. He merely asserts that the artefacts we call knowledge are necessary stand-ins, imperfect but indispensable for the purposes of communication, storage, and retrieval. The transformation from inert record to lived understanding happens beyond the document, but that does not negate the value of the document in facilitating that process. In this way, Dillon navigates a middle ground: he retains Wilson’s epistemological caution but bypasses it in favour of practical system design. When KM makes knowledge manageable it results in a focus on codifiable artefacts.

If we reject the hierarchical interpretation of the tetrarch DIKW model, we come to understand that knowledge is not located at the midpoint on a stable ladder, but rather in a domain of fluid, negotiated interplay. In this light, attempts to manage knowledge as a singular, ownable entity begin to appear both conceptually fragile and politically charged.

If knowledge is rendered manageable through the DIKW model, how do we ensure that we are not only asking what can be codified, but also whose knowledge is being privileged and for what purpose?

Next part: Section 2.6 – The will to metrics: KM’s economic telos.

Article source: Koskinen, H. M. (2025). The Emerging Concept of Responsible Knowledge Management (rKM): Identifying and Formulating the Core Principles of rKM. (Master’s Thesis, LUT University).

Header image source: Created by Hanna M. Koskinen using ChatGPT.

References: