Enabling an ecological model, preferably with intelligence

In the second of a two-part series Dave Snowden identifies some of the principles that characterise a naturalistic approach to knowledge management and explains why ‘recipe books’ are no longer useful

In part one, ‘Breaking the engineering paradigm’ (see Inside Knowledge, August 2010), I used an s-curve model to show the progression of ideas in management from scientific management, by way of systems dynamics to cognitive complexity. I argued that while the engineering models that dominated period from the 1980s to the current day had (and continue to have) considerable utility, we have reached the limits of their applicability, and in many cases (sick stigma, for example) taken them to access to the detriment of human intelligence and innovative capacity. In order to use technology a symbiotic augmentation of human intelligence, we need to break with the structured and directive thinking, and practice, that characterised both scientific management and systems dynamics, then shift to approaches based on the natural sciences. This means working with the way humans have evolved to be, rather than trying to engineer an idealised and homogenised model of either practice (BPR, 6?) or values (5th Discipline et al)1. That involves changes not only in what we do in KM, but also in how we think about the subject.

Changes in thinking
Shift from emphasising order to enabling messy coherence
One of the best teaching stories I have ever created is the Children’s Party Story2. You don’t manage a party with formal objectives, motivational video-tapes and key performance indicators (KPIs). Rather, any sensible adult sets some boundaries (generally flexible and thus maintainable and capable of being monitored) then provides various options for play. Good emergent patterns are then reinforced, bad ones rapidly disrupted. The adult here does not micro manage, rather they manage the emergence of beneficial patterns. They do not assume that they can know in advance what will be good, but they can recognise it when they see it.

Humans work best in a cycle of order and chaos. Think of your own work area; for most of us, it gradually descends into incoherence before we suddenly indulge in a a tidy-up session, creating order which then starts the gradual decline into mess. The reason we do this is that it works. If we kept rigidly to the structure we put into place last January then we would not be adapting to change, we would be attempting to ignore it. Imposing a rigid structure for highly stable information sources (engineering design documents, for example) is a good idea; using the same approach for human nous or canniness3 is an over-constraint that prevents new ideas and practices from emerging.

Stop creating and using recipes – instead, educate chefs
If I don’t know anything about cooking then I can do a lot with a recipe book. Assembling the ingredients and following the step-by-step instructions will generally produce an acceptable menu. The problem is that when I don’t have the right ingredients or the right environment life gets difficult. Employ a recipe book user to cook a meal for you and it will probably work but you will need to allow time for ingredient assembly, reading of the book and the step-by-step following of instructions. As the meal gets more and more complex, multiple recipe books will be open in the kitchen and life can start to get interesting. As complexity increases information overload sets in and things go wrong. Worse still if several people are involved in the meal and things go wrong then sooner or later they will start to allocate blame based on failed process.

Now let’s contrast this with a chef. A good chef may have started with recipe books, but s/he also served an apprenticeship doing basic tasks in support of the master, observing practice and swapping stories with other apprentices. As s/he advanced in the art they were trusted with more complex tasks. Asked to provide a meal a chef is able to accommodate for the environment in which you are situated, using available ingredients. If things go wrong then they can adapt quickly to changed circumstances. Above all they can manage parallel activities, their knowledge is embodied, it is a part of what they are. Aristotle understood this many years ago when he talked about sophia (the ability to reflect) and phronesis (the ability to act or practical wisdom). Both are needed.

Probably the major error of the engineering period in knowledge management (KM) has been the assumption that it’s all about creating recipe books; something that needs to be urgently addressed. It’s not just the embodied knowledge by the way, it’s also about the ethics of attention, something that is drummed into you as an apprentice and has to be experienced. It cannot be learnt.

Shift from models to frameworks
The two terms are often used as synonyms, but there are some fundamental differences. A model seeks to create a representation that will allow prediction of system behaviour. In contrast, a framework provides a way of looking at the system from different perspectives. Highly complex human systems cannot be fully modelled. Indeed, Gell-Mann and others have argued that the only valid model of such a system is the system itself. Frameworks such as Cynefin4 provide different lens or perspectives from which a situation can be viewed. If the system is simply engineer it to death, apply best practice. If it’s complicated then bring in the experts or carry out an investigation to discover options for good practice. If complex (the children’s party) set up safe-fail experiments and see what works, allow for emergent practice. If you are desperate remove all constraints to create a chaotic environment in which novel practice can emerge. A good framework does not legitimise one universal practice, but it does create the boundaries within which different (and sometimes contradictory) approaches can be sensibly adopted.

This shift from fail-safe design (deciding what is an ideal future state and seeking to achieve it), to safe-fail experimentation (interacting with the ecology to discover what is sustainable) is probably the major strategic shift of the current age. KM people should understand its necessity and ideally should lead practice here.

Heuristics of a naturalised approach to KM
Many years ago, I formulated three basic rules for KM, which I gradually expanded to seven. In a slightly updated form, they now comprise a guide to practice.

1. Knowledge can only be volunteered. It cannot be conscripted
You can’t make someone share their knowledge, because you can never measure if they have. You can measure information transfer or process compliance, but you can’t determine if someone has truly passed on all their experience or knowledge. The implications of this are that systems which link and connect people to each other will be more effective for one simple reason: people understand the context of a question better than a search engine or hierarchical taxonomy. Social computing products such as Twitter allow people to ask questions and if people find you interesting and trustworthy then (and only then) do you get answers. Many years ago I developed an open-source technique called social network stimulation5 which does not manage knowledge, it aims to reduce the degrees of separate between employees. By building a network you enable knowledge flow, and investing in plumbing is better and more sustainable than investing in content.

2. We only know what we know when we need to know it
Human knowledge is deeply contextual and requires stimulus for recall. Unlike computers we do not have a list-all function. Small verbal or non-verbal clues can provide those ah-ha moments when a memory or series of memories are suddenly recalled, in context to enable us to act. When we sleep on things we are engaged in a complex organic form of knowledge recall and creation; in contrast, a computer would need to be rebooted. In terms of practice this picks up on the earlier point. Bringing people together from different backgrounds to look at a problem increases the scanning range of the group; more and deeper memories are stimulated. Workshop techniques such as ritual dissent6 increase the capacity of a group to go beyond entrained thinking

3. In the context of real need, few people will withhold their knowledge
A genuine request for help is not often refused unless there is literally no time or a previous history of distrust. On the other hand ask people to codify all that they know in advance of a contextual enquiry and it will be refused (in practice its impossible anyway). Again, the same basic theme of investing in linkages between people to allow content to flow. In some of the KM systems I have designed or advised on over the years, I generally recommend that people focus on sharing metadata not data. This not only is cheaper to set up, but it also means that people are not being asked to share the data itself, just data about the data. That means that permission for use can be given at the time in context, rather than the impossible task of securing permission in advance without context.

4. Everything is fragmented
We evolved to handle unstructured, fragmented, finely-grained information objects, not highly structured documents. People will spend hours on the internet, or in casual conversation without any incentive or pressure. However creating and using structured documents requires considerably more effort and time. Our brains evolved to handle fragmented patterns not information. Social computing succeeds because it utilises fragmented not highly structured material. One of things my own company developed is an approach to KM based on capturing thousands of self-indexed micro-narratives (something very different from storytelling) in the field under fire (literally in the case of one client). Faced with a complex problem, a field engineer (or any other knowledge worker for that matter) does not want an idealised best-practice document. They want immediate access to the experiences of other people in similar situations. For years at KM conferences I have asked a simple question: given an intractable problem would you prefer to hear ten or fifteen stories of people in comparable situations, or read a best-practice document. To-date no one has opted for the second one, despite it being the focus of most KM practice. Pictures, field notes, impressions are all as (if not more) important as structured material.

5. Tolerated failure imprints learning better than success.
When my young son burnt his finger on a match he learnt more about the dangers of fire than any amount of parental instruction could provide. All human cultures have developed forms that allow stories of failure to spread without attribution of blame. Avoidance of failure has greater evolutionary advantage than imitation of success. It follows that attempting to impose best-practice systems is flying in the face of over 100,000 years of evolution that says it is a bad thing. That means a focus on worst practice systems is as important as best practice. The sort of micro-narrative systems referenced above do this well; sharing an anecdote of failure is very different from confessing failure in a formal document. Other techniques, such as the use of archetypes, also have a long history of success in human society. Telling a story about something that an archetype did enables confession of failure without attribution of blame. What won’t work under any circumstances is a pious statement to the effect that we have a no blame culture. Most employees have heard that one too many times before.

6. The way we know things is not the way we report we know things.
There is an increasing body of research data which indicates that in the practice of knowledge people use heuristics, past pattern matching and extrapolation to make decisions, coupled with complex blending of ideas and experiences that takes place in nanoseconds. Asked to describe how they made a decision after the event they will tend to provide a more structured process-oriented approach which does not match reality. This has major consequences for KM practice. In particular it argues against systems based purely on reflection of what happened and more use of ethnographic techniques, capture at the time of immediate impressions and more generally about more advanced techniques such as formulating heuristics, the use of ritual and metaphor as a communication technique. Again in projects we have improved health and safety compliance by creating clothing-based rituals to trigger different cognitive response patterns. This is far more effective that rules and short training courses. In one advanced research and development project, I am looking at metaphor-based languages to communicate complex ideas. A one or two-word reference to a commonly understood situation is the natural way of passing on expert knowledge. ‘Do you remember when we were in X, we did Y?’ is a common form of communication between experts and this can be formalised effectively and incorporated into training programmes

7. We always know more than we can say, and we will always say more than we can write down.
This is probably the most important. The process of taking things from our heads, to our mouths (speaking it) to our hands (writing it down) involves loss of content and context. It is always less than it could have been as it is increasingly codified. It is derived from Polanyi and shows the basic error in Nonaka’s SECI model, which has informed so much KM practice over the years. There are things we can codify in written form, which has been the legitimate focus of KM practice to date. However it is very limited. The narrative approaches we have pioneered address what can be spoken or codified in symbolic form which adds considerably to the richness and utility of a system.

What we know through a combination of practice, education and reflection, is that the intuitive understanding of a chef should be the real focus of KM. Recipe books are no longer good enough.

Dave Snowden is the chief scientific officer and founder of Cognitive Edge Pty Ltd. He can be contacted at dave.snowden@cognitive-edge.com

References

1. For those wishing for more depth than is possible in this short article I have two chapters (one with Gary Klein) in the recently published Informed by Knowledge edited by Mosier & Fischer which elaborate on these and other ideas;
2. To hear the full story, visit www.youtube.com/watch?v+Miwb92eZaJg;
3. A wonderful word of Scottish origin which means shrewdness combined with good judgement;
4. See “A Leaders Guide to Decision Making” Snowden & Boone, HBR Nov 2007 and specifically applied to KM at www.cognitive-edge.com/ceresources/articles/13_Complex_Acts_of_Knowing;
5. Method available at http://www.cognitive-edge.com/method.php?mid=43;
6. Method available here http://www.cognitive-edge.com/method.php?mid=46.