The nature of knowledge

Exploring an alternative knowledge architecture

For all our attempts to manage it the nature of knowledge remains elusive. Simon Hudson reflects on traditional attempts to define knowledge and presents an alternative approach to designing systems to help store and navigate knowledge rather than data.

In designing knowledge-centric systems I have been struck by certain observable facets of knowledge facets that are intriguing and deeply complex.

APCQ defines knowledge management as: “A conscious strategy for getting the right information to the right people at the right time and helping people share and put information into action that will improve organisational performance.”

It is the nature of knowledge itself that proves elusive. The temptation is to follow one definition with a string of further definitions and let the reader decide. This ultimately adds little value. Individuals tend to select the option most closely matching their existing perceptions. Alternatively one might claim the mantle of a KM guru and espouse our own definition(s). This I shall also resist despite the vicarious temptation it offers. Instead in this article I will draw on an approach often used in science by offering a model for how knowledge might be represented irrespective of what it actually is. I will take the relationship with science further however and consider some parallels between what might be considered ‘old’ and ‘new’ in both science and knowledge management.

A thought experiment

To start let us set up a thought experiment. Imagine a hypothetical knowledge store that holds all the accumulated data information knowledge and wisdom of mankind.

Now the ‘traditional’ way of managing this might be to treat it all simply as data which we would propose to convert to more useful information by the process of rigid organisation and categorisation. This is the stuff of IT systems and classic databases. It operates on the presumption that everything about the information can be defined and that the relationships between items of information can be understood predefined and crucially is consistent irrespective of the use or the user. We might imagine millions of data tables; containing terabytes of data in billions of records each. These tables would be linked to each other using millions more predefined relationships each encapsulated in a huge SQL statement or something similar.

We might further imagine this totality of human knowledge being accessed through hundreds of thousands of information terminals in each country using predefined reporting tools and data-mining applications able to navigate the neat highways of this knowledge.

Who among you thinks this a laudable achievement perhaps the pinnacle of human know-how?

I would like to believe only a minority would answer in the affirmative for this is the information vision of early 20th century science fiction perhaps Asimov’s Multivac supercomputer and it is a fundamentally flawed image. The fallacy lies not in the vision but in the understanding of the nature of the relationships in the data.

Such a model might be considered to be highly Newtonian by a 20th century physicist in that it works well within certain limits and is capable of being applied to narrow and deterministic situations to allow good predictions to be made. It assumes mistakenly that the user of the information does not affect the nature of the knowledge itself.

Fortunately the universe isn’t so dull and has a delightful capacity for confounding such simple (sic) models. I turn to a favoured quotation: “For every complex problem there is a solution that is neat simple and wrong.”

Non-deterministic knowledge

True knowledge has two complicating features which I shall endeavour to describe in terms familiar to physicists.

First the nature of the information upon which knowledge is founded is strongly affected by the user/observer. This comes about because he or she interprets any information in response to existing experiences other knowledge and with a view to the intended or subconscious application of the information. In this respect it is unlikely that everything about the information can be predefined and relationships between the data tables in our earlier thought experiment would depend not on the data but on the user/observer. Here we see the Newtonian model begin to fail. It hints at the power of knowledge being not in what is but rather in what individuals make of it (both figuratively and literally). This feels much like a form of special relativity for knowledge though sadly lacking in the mathematical rigour to more properly describe it.

This user/observer effect gives rise to a second phenomenon that of granularity and chaos within collections of knowledge. Since the relationships between items of knowledge are so personal and rich it is possible to delve into the network of connections between our items of knowledge and discover deeper layers of relationship hidden within those we initially saw; perhaps more subtle than the last. At times trying to link such information would produce apparent chaos while elsewhere in this hypothetical knowledge store would be huge clusters of neatly structured data (enough to gladden the heart of any Newtonian IT developer).

This is the fractal nature of knowledge in our thought experiment. Imagine our global knowledge store as one would look down on a planet then zoom in on one ‘continent’ of information that interests you then pick a particular country and within that a feature of interest say a piece of coastline. As you draw closer the complex outline of the coast gives way to equal complexity until you are looking at individual grains of sand. Beyond lies the complex beauty of crystal structures complex molecules and down through the sub-atomic world to the undiscovered country beyond. Now do all this again but in rather more dimensions than the simple three visualised.

At times what you see might seem to make sense other times it might remind you of other patterns you recognise from elsewhere while again it might seem to be totally random chaotic or meaningless. Whichever direction you drill through this information higher lower left or right in or out the effect is repeated. And remember that a different observer sees the patterns different to you. I imagine it may be possible to mathematically describe a fractal coefficient for certain types of information implying the degree of sensitivity of interpretation to the user.

Applying non-Newtonian approaches

So how does this model help us deal with the nature of knowledge? First let us not abandon the Newtonian (traditional IT) approach – it continues to yield massive value. Newtonian mechanics survived for hundreds of years providing predictions at better than 99 per cent accuracy for almost everything it was applied to; the mistake was to apply its rules to everything. So don’t assume that the traditional methods can be used for creating information stores for all applications. The act of forcing knowledge into a neatly codified and categorised knowledge store often has the effect of degrading the richness of information itself by eliminating much of the granularity of its connection with other knowledge.

Forcing a square peg into a round hole rarely leaves you with a square peg.

Somehow the knowledge needs to be stored in such a way that its richness can be exploited. This means using flexible expandable and evolvable storage architectures in place of rigid ones. It means not relying on information technology alone although IT has a part to play. It means building information systems and services around the people who are to use the knowledge anticipating their most likely uses for the knowledge anticipating the tendency for these to evolve over time and even alter radically at times (and not getting upset or defensive when they do). It means exploiting people’s inherent agility at manipulating knowledge and helping put people in a physical and cultural environment where this can be maximised while creating processes for capturing their learning and insights for possible re-use.

I propose therefore that any knowledge storage system needs first to consider whether the information to be contained is narrow in scope and readily specifiable – accounting information for example. (If my earlier proposal that a coefficient might be established then this would be low coefficient information.) If so then feel free to be as Newtonian as you wish. If not then a different architecture would be more effective. I foresee this new architecture embodying many of the concepts routinely discussed in this journal but also one that would promote the creation of links between documents and other forms of information including other knowledge systems (even the Newtonian ones). These links would be non-exclusively categorised and identified as belonging to individuals groups and populations. In many respects the linkages are as important as the information they connect and so deserve a similar degree of meta-data to enrich them.

A minimal example of this is common on certain e-commerce sites where having selected a particular book or compact disc you are offered a selection of other titles that people who have purchased your selection also bought. If we were to apply this to our knowledge system we would have the facility to follow others’ thinking via a self-creating and automatic knowledge-map or continue to press on (via other knowledge retrieval methods) to create new pathways that future explorers might examine. Extensions might include:

Making it obvious which are the more commonly followed routes (thicker lines in a mind map);
Resolving other individual’s pathways from the aggregated set linked to a profile of that person’s (professional?) interests and skills and linked in turn to other research knowledge-maps they have pursued;
Direction of travel between documents duration of stay at the target document.

At last our system begins to reflect and sustain the profound nature of knowledge. We can use it to explore the information itself or its relationship with other knowledge sources or even something about other users of the information and where it may have taken them. It would be interesting to see how the evolving patterns of linkage compared with any fractal coefficient that had been derived.

Conclusion

‘Information overload’ has become a truism of the information age. If we are as a (increasingly global) society and as individuals to harness the power that effortless and ubiquitous access promises then we need to learn new means of manipulating that information in order for it to become knowledge rather than noise. We may consider information overload as a symptom of the immaturity of our approach thus far. In our haste to exploit our new found information tools we have not delved deeply enough into how we gain knowledge from information. We have not appreciated its exquisite subtleties and we have underestimated the impact that the user of the information has. The nature of knowledge remains mysterious but we are perhaps beginning to map the borders of our ignorance.

As the industrial age heralded only the beginning of engineering and manufacturing accomplishment and ultimately gave us the tools of the information age so we may expect to see much development in this the nascent years of the knowledge age.

(c) Simon Hudson 2001

Simon Hudson is e-marketing and strategic intelligence manager at Smith & Nephew. He can be contacted at: simon.hudson@smith-nephew.com

The walled city – The role of knowledge ‘gatekeepers’

We might wish by way of setting the scene to think back to medieval times. It was common for cities to be surrounded by gated defensive walls; gatekeepers were used to keep out thieves and other undesirables. In principle this was a sound approach. However it routinely resulted in long lines of people waiting to enter the city including dignitaries traders and other people of intrinsic value to the running and commercial wellbeing of the city. The gatekeepers themselves were rarely popular and there is little evidence that cities so equipped were any less free of crime than those without. Indeed it seems obvious knowing what we do of the type of person employed to such a task that a city gatekeeper would have had limited ability to differentiate the thieves from the honest merchants and almost certainly little incentive. Furthermore any city that fooled itself into believing the gatekeeper to be effective and so relied solely on them would have had no mechanism for dealing with problems or disruption within the city walls. This latter role was commonly associated with the town guard or city watch who would roam the streets preventing and correcting trouble. Cities with a well-established watch would have been far more accessible and profitable as trade was able to flow more freely.

It is my belief that this is a strong metaphor for the security component of a knowledge management intranet. I ask you to seriously question the value of gatekeepers anywhere where the free exchange of information is desired. If the analogy above holds true then the use of gatekeepers certainly holds up the flow of information into and out of your knowledge sharing system or intranet while they would seem to do little to eliminate poor quality information. Also does it not create a counteractive culture where the positive behaviours of those attempting to share knowledge are first met by steps to prevent them?

Instead I offer the use of the town guard approach to police the content acting on transgressions in a firm but fair way. Such ‘knowledge police’ should not create an aura of paranoia but rather be part of the community (much as the idealised British bobby would); offering support from a position of knowledge and respect. They should be visible and empowered and required to promptly take action. They should help people into the ‘knowledge city’ show them around and point out what is and is not permitted. Equally it should be the duty of everyone within the knowledge community to note undesirable activities and either tackle it themselves or bring it to the attention of those charged with maintaining order. Peer pressure can be a powerful force within any community.

Suddenly we have everyone contributing to the wellbeing of the knowledge community and being conscious of the need to maintain good behaviour without placing draconian constraints on the flow of knowledge within or across the boundaries of the community or system.

Before the question is raised let me state that I do not advocate tearing down the city walls – every community or asset needs a form of security. But it seems that those within the community using the knowledge infrastructure are well placed to monitor the activity of the citizens they share the space with.

Final thoughts

It has been my experience that it is sufficiently difficult trying to create a knowledge sharing culture within an organisation to invite people to set up residence within the city as it were. Hence any form of barrier to this may be considered a ‘bad thing’ if the ambition is to create such a culture. Gatekeepers are by design a powerful barrier both physically and psychologically. While some parts of the city may need to be secured with such a potent instrument the more successful ones will surely avoid overuse; perhaps diverting the resources saved into knowledge ambassadors instead.

By Simon Hudson