Knowledge Management: a position paper


Piero Pagliani
Research Group on Knowledge and Communication models
  1. Introduction
  2. The economic scenario
  3. New management, decision making and Knowledge Management
  4. Determination problems and choice problems
  5. The det-choice approach
  6. Enactment, open systems and decision making
  7. The role of Knowledge Management
  8. Conclusions
  9. Coming soon. From speed optimisation to depth optimisation: overcoming Taylorism by means of the Organization Semiotic Optimisation Approach
Click on the barlines to come back to the table of contents


1. Introduction

Although I come from the Artificial Intelligence business, I retain that Knowledge Management has only an indirect relation with AI. For example, Office Automation problems exhibit features which, for KM, are not less relevant than AI.
Nevertheless, I think that AI should gain a more recognised citizenship in KM, and this is possible if we look at it from a somewhat forgotten point of view; a point of view which, on the contrary, was one of the most important motivations of AI researches. I am thinking about the streamline started by H. Simon's issues about traditional utilitaristic (or neo-classical) foundations of Economic Theory.
As is well-known, according to Keynes, H. Simon rejected the axiom of "complete comparability" of rational choices and that of "transitivity of preferences", substituting them with a theory of "limited cognitive capabilities", or "limited rationality".
In this framework, "intentionality" cannot be understood outside a contextualised (or "situated", we would say nowadays) meaning attribution.
The new keyword was heuristic, to be understood as a tool for acting with limited resources in limited informed situations.
It follows that knowledge-making and decision-making depend also on "irrational" parameters, such as emotions, common visions, social contexts, personal and collective beliefs, personal and collective sensemaking processes (1).


2. The economic scenario

When goods were characterised by long life-cycles, that is, when technologies were used to produce the same goods for long intervals of times, the "needed knowledge", for any production process, was embedded (at least in principle) in firms using the hierarchical templates of work organisation, so that it was transmitted and received by pouring it through the given hierarchical channels. Any new knowledge was usually achieved through planned "education programs" and class-room teaching, well defined both from a physical and a temporal point of view.

This was possible because a firm was thought of as a "closed system", where any interaction with the external world was planned in advance. In turn, this was justifiable because the customer-supplier cycle was long in nature and, finally, this was made possible by the technological characteristics of the products and by the relative immaturity of the market (it seems that Ford used to say "People want black cars because I produce only black cars").

On the contrary, nowadays we are living in an advanced stage of a migration process from the above scenario, sometimes characterised by the term "mass production" (production dominates market) to a new one called of "mass customisation" (the reverse of the former) in which life-cycles of goods and services and their technological evolutions, are subjected to a strong acceleration, general and new in the industrial history.

From a purely economic point of view, we are facing the following connected phenomena:

1) the stability in time of goods decreases, while the differentiation of the production factors increases;
2) the functional dependence between "production process efficiency" and "production process value" decreases, while the dependence between production factors increases.


3. New management, decision making and knowledge management

The loose connection between efficiency and value together with the tight connection between different production processes, makes a reliable computation of the marginal effects of economic initiatives almost impossible. Economic "predictions" become more and more subjective "expectations" and decisions (for instance investment decisions) rely less and less on some form of "objective" or "rational" evidence.
Under another perspective, we can say that the "mass customisation era" is characterised by open (and distributed) systems and not by closed systems as used to happen before.

Because of the intrinsic complexity and the intrinsic dynamics of open distributed systems, we can hardly now view decisions as means for implementing a meaningful interpretation of the world.
On the contrary, sometimes they are now described as means for making meaningful interpretations of worlds which basically are confused among themselves (2).

Moreover, explicit and coded knowledge becomes less and less effective, whereas management has a less and less precise idea of what explicit and coded knowledge is needed in order to maintain a sustainable competition in time.
Indeed, myths, metaphors, redundancy and tacit knowledge become fundamental tools in organisations, as we have learned, for instance, from the best Japanese experiences.
Preferences, beliefs and knowledge, on the one hand, together with problems and solutions, on the other hand, are interactively constructed by means of enactment and sensemaking processes.
Thereafter, we cannot regard information, meaning and knowledge as something that is objectively embedded in a code that a receiver shares with a source, any longer. On the contrary, we basically do not know the "code" used in the message, for the simple reason that the interpretation conditions, the sender context, the receiver context and so on, change continuously so that any "message" must be reconstructed interactively (whatever the source-receiver link is: outside world-human, human-computer, human-human, me-you, me-me).

Therefore, the structuralistic pouring model

has to be transposed into the enacting model

(that, incidentally, is coherent with Peirce's interpretation theory), in which interpretation is a dynamically contextualised constructive act.

The former model parallels the rigidity of "mass product-oriented" work organisations, the limited variety of their products, their vertical divisional structures and the predetermined form of their information flows. On the contrary the latter is conceptually connected to network-based organisations with flattened hierarchical levels and co-operative forms of knowledge creation and circulation. A kind of organisation that is more reactive and more "mass customisation-oriented".


4. Determination problems and choice problems

The enacting model could be formalised in a few steps, starting from the following diagrams -see (3):

As Lawvere and Schanuel explain in (4), finding raw d (in the left diagram) is a "determination problem": roughly speaking, I have a process q with input A and output C and I have to determine a sub-process d that after a sub-process s with input A and partial output B, makes it possible to complete the transformation of A into C. Applying recursively this procedure we obtain a top-down and forward "factorisation" of our original process d. This is exactly the procedure usually applied in the "mass product" era.
However, nowadays we face the problem illustrated by the right diagram above: I start from the output C (because C is now conceptually an input, not an output. Namely it is the input given by the contextual space, for instance the market) and I must backtrack to understand what I eventually need before the sub-processes recursively linked to the conceptual input C, say d, in order to obtain a hypothetical process that transforms a resource A into C. In this way I achieve a top-down and backward reconstruction of the needed process q. According to Lawvere and Schanuel, this is a "choice problem".

Here "top-down" means that we still apply a "from general to particular" approach. Nevertheless, the point of view drastically changes in the backward approach (5).
However, we will see below that a two-sided approach is actually unavoidable.
In fact, as soon as one recognises the existence and the relevance of the choice process, one has to note that choice problems and determination problems define each other. There is no notion of "choice" without a notion of "determination" and vice-versa (
6).


5. The det-choice approach

The interdefinition between choice and determination, induces a backward-forward double and syncronic analysis (that we can shortly call "det-choice" approach) that cannot be seen as a traditional top-down approach. In fact the top-down methodologies are characterised by the functionalistic point of view, according to which it is possible to analyse a complex entity (such as a complex problem or a complex process) in terms of atomic constituents on which the given entity functionally depends. These functional components are achieved through a one-way breaking down process.
On the contrary, I surmise that the det-choice approach here proposed is not really a top-down methodology, because its double movement does not single out atomic functional components, but approximations of atomic components, achieved by means of a sort of negotiation between determination factors and choice factors.

Now, let us see a possible pre-formalization of this approach.
From a conceptual point of view, d is what I need to get q after a given s. That is, an implication from left (before) to right (after): d=sàq. Dually, s is what we need before a given d to obtain q . Hence it is an implication from right (after) to left (before): s=qßd.
(In a sense, it is this "coming back from the future" that neutralises the rational and deterministic, although unreliable, forecast of marginal effects).
Henceforth, what we intend as a meaningful world is both determined and chosen:

Observe that C is not a mere target, but rather a common space of interaction that mediates determination and choice issues.
This two-sided process is what we call det-choice.
From the discussion above, one can straightforwardly verify that it is an implementation of the notion of "enactment".
It is absolutely an approach which is dynamic in nature, since in the meanwhile it requires and determines some milestone, some semantical stipulation, some "formal ontology", some common measure, in short some common feature to be used as a "meeting point".
It should be clear that we are not speaking of the well-known milestones used in some problem solving algorithms of classical AI. In AI milestones are fixed in advance. In our case, as we have already noticed, we are speaking of an approximation process, where such an approximation is obtained by means of the det-choice process, à C ß, itself.
The det-choice technique has been first applied (and implemented) for approximating relations by means of relations, in
7 and 8.
In spite of the specific mathematical context, these applications have a much more general meaning, since relations are used to model both processes and patterns of resources.

The det-choice approach does not imply that real world must be conceived as a subjective construction, but, according to F. Varela, as a hammer beating on a piano keyboard with a constant and uniform movement. On the contrary, the piano itself can freely move to the right and the left, at different speed. It is this subjective movement that creates music. That is to say, we are not "played by the external world", but we and the external world play together (9).


6. Enactment, open systems and decision making

In accordance with the enactment philosophy, decision making is more an interpretative act than a rational choice, as was observed in the specialised literature (10). Henceforth, since language, symbols, common vision and social interaction are part of any organisation's sensemaking, they are part of any decision, implicitly or explicitly.
"Organisations are communication and ambiguous action networks, in which solution gaps are filled mainly with a system of reciprocal commitments, having fuzzy and variable contours. Fixed routines and formal roles are minor ingredients." (11).
"And", we could add, "in which explicit knowledge gaps are filled with tacit knowledge", that is a knowledge that does not respect the organisation hierarchies and their parallel organised information flows.


7. The role of Knowledge Management

If we accept the above conceptual framework, then KM with its capability to collect and organise, to some extent, tacit knowledge and to mix it with explicit and formal knowledge, should be seen as a tool for implementing the famous "plausible reasoning" required by H. Simon's "limited rationality" hypothesis. In other words, KM should be a tool, or rather a system of different tools, for supporting that weakened form of "rational reasoning" which is required in such a fluctuating, atomised and parcelled framework.
The "top-down (one-way) philosophy", so old and so venerable, and its related tree-like organisation discipline, must then be substituted by an "actor-based" philosophy according to which, (almost) autonomous actors interact among themselves in many ways, in order to make views, to associate in temporary teams, to react and pro-act in the environment.


8. Conclusions

This transition from a transformational setting to a reactive one, is made dramatic and pressing by the globalisation process.
KM should enable an organisation to move from a solid state to a fluid state, without dissolving. This is required because, in actual Economics, solid mechanics (and its venerable equilibrium theory) has to be substituted by fluid mechanics (and its aggregation theory), or even by Quantum Mechanics (with all its unusual and un-intuitive principles, like the packet-wave contradiction, the wave function, the uncertainty principle, the participation of the observer in the observed phenomena, the concepts of resonance and duality (12)).
To this end, modern Logic and Computer Science have a number of models to be borrowed by Management Science (for instance, systems with various local kinds of logic, unifying logical systems, distributed logical modules, object-oriented and declarative programming paradigms, actor theory) (13).
But also the reverse side is true (14).
However, things are not so easy and so clear. To some extent, the relation between a transformational paradigm and a reactive one, is the same as the relation between the Turing machine paradigm and a new, required but still unknown, computing paradigm based on the very notion of "interaction".
Despite actual progress and exaggerated (commercial) vaunting, we have just started the job: just as we still need a new mathematics and a new logic in order to achieve a new computing paradigm, we still need new conceptual co-ordinates in order to frame Organisations into the interactive paradigm.
Presently, in both cases people apply good practices within an empirical and pre-methodological work in progress.
The Internet works anyway and the event-driven, multitask, Operating System of your PC too. That's fine. From time to time our computer suffers "internal conflicts" and "internal errors" induced by an heterogeneous and a slightly anarchic application software complexity. We can tolerate them. But we cannot tolerate low speed interactions and unpredictable response-time, because this induces (a privileged form of) anxiety.
In sum, there is still much work to do, both theoretical and practical.
For the moment, we can just tell "what does not work any longer, what we do not want any more".
 

Any comment or question about the ideas and concept introduced in this paper, are welcome

Home page


 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 


NOTES

(1). "[...] while the arguments of traditional epistemology focus on "truthfulness" as the essential attribute of knowledge, for present purposes it is important to consider knowledge as a personal "belief", and emphasise the importance of "justification" of subjective knowledge. This difference introduces another critical distinction between the view of knowledge of traditional epistemology and that of the theory of knowledge creation. While the former naturally emphasises the absolute, static, and non-human nature of knowledge, typically expressed in propositional formulas in formal logic, the latter sees knowledge as a dynamic human process of justifying subjective and personal belief as a part of aspiration for the 'truth'." (I. Nonaka, On Knowledge creating organization. In Atti del XII Convegno Nazionale AIF: Nuovi Alfabeti. Linguaggi e percorsi per ripensare la formazione. Parma, 1993).

(2). J. G. March, "A primer on Decision Making. How Decisions Happen", The Free Press, New York, 1994, page 189

(3). P. Pagliani, Knowledge and Knowledge Management (in Italian), "Prometheus" (in print).

(4). F. W. Lawvere, S. H. Schanuel, "Teoria delle Categorie". Franco Muzzio, 1994.

(5). One could apply a more radical inversion, namely a "bottom-up" approach (that is, "from particular to general"). I think that this is too dangerous, at least for medium-large organisations, since it risks curing an organisational atrophy by means of an organisational anomie.

(6). This dialectic "determinism-choice" is absolutely general. In fact, notice that a decision is really a choice when it is linked to deterministic effects while it is not a choice when it induces random effects. In this case I do not choose anything.

(7). P. Pagliani, Modalizing Relations by means of Relations: a general framework for two basic approaches to Knowledge Discovery in Database. In "Proc. of the International Conference on Information Processing and Management of Uncertainty in Knowledge-Based Systems. IPMU '98", July, 6-10, 1998. Paris, France, pp. 1175-1182.

(8). P. Pagliani, "A Practical Introduction to the Modal-Relational Approach to Approximation Spaces". In L. Polkowski, A. Skowron (eds.): Rough Sets in Knowledge Discovery 1, Physica-Verlag, 1998, Ch. 11, pp. 209-232.

(9). F. Varela, Il reincanto del concreto. In P. L. Capucci (a cura di) "Il corpo tecnologico", Bologna, Baskerville, 1994, pp. 143-159.

(10). K. E. Weick, "Sensemaking in Organization". Beverly Hills, Sage, 1995, page 185.

(11). L. Biggiero & D. Laise, Determinism and intentionality in some research traditions in Economics and Management. (In Italian). In M. Cini (ed.) "Caso, Necessità, Libertà". CUEN, 1998, page 146 (my translation).

(12). Isn't a modern Company a "packet" qua connected and individualised body, but a "wave" qua organism continuously pervading the external world and continuously pervaded by it? Isn't the active participation of both customers and suppliers in the value chain of the Company, an example of participation of the observer in the observed phenomena?

(13). "In the "transformational" paradigm, the system reads an input, processes it and produces an output. The input is here the initial query and the output is any possible answer context C. [...] In the "reactive" paradigm, several agents interact together by exchanging messages. There is no notion of input and output in this case; the initial query is used only for the purpose of bringing into life certain agents. The answer context C acts as a medium of communication between agents." ( J. M. Andreoli, R. Pareschi: Communication as Fair Distribution of Knowledge. In "Proc. of OOPSLA'91", Phoenix, USA, 1991).

(14). First, let us observe that an organization is characterized by essentially two kinds of behaviours: internal distribution [...], external cooperation [...] . The distinction about the two operations available for an organization to handle different kinds of tasks gives us an interesting insight on the sharing of knowledge in object-oriented systems." (J. M. Andreoli, R. Pareschi , LO and behold! concurrent structured processes. In "Proc. of OOPSLA'90", Ottawa, Canada, 1990).

Top