I’m bringing this series to an end in a storm-blown cottage in North Wales. All part of training for the Annapurna circuit at the end of March with Daughter. We booked the week to coincide with some work at Bangor and my vision of the week was clearing work early morning and evening with 5-7 hours on the hills with winter sun glinting on snow. The reality of today’s brief walk was torrential train and a wind strong enough to threaten stability in the valley let alone the tops; so we did an early exit from a walk which was exhilarating to say the least. The iPhone is now in the rice container in the hope it recovers; the water penetrated even the waterproof pocket of the winter jacket. Writing this blog had made me more aware of the role of experience in exercising judgement. Crossing a mountain stream between down blasts, picking a route to reduce the risk of being blown over, knowing when the time had come to take an early exit; all were more or less instinctive decisions derived from years of walking and a lot of stories heard and retold in pubs at the end of the day. For that type of knowledge, there is no substitute.
So to conclude this series I wanted to return to the three types of knowledge I identified in the first post and briefly summarise how the approach I have advocated addresses them. I’m not going to link to prior posts or links as you really need to read the series for this one. Remember when I started this series I said (of the issues listed below): All of those require approaches that are far from conventional in nature. In devising them we can be drawn on one of the most ancient, and most effective, knowledge transfer systems namely the apprentice model. We can use digital means to enhance that, but not replace it. That picks up with a general theme I have been developing over the years, namely that technology should augment, but cannot replace human intelligence.
I have replicated words from the first post in italics as a quote and then proved a summary of how I have handled the need in previous posts.
A large part of experience is abductive in nature, the ability to link or connect what to the less experienced is a disconnect. The more memories you have access to, the more diverse the experience that gave rise to them, the more extensive the networks which extend them, the more you can make those links under pressure.
One of the principles of knowledge management I created a few years ago was: We only know what we know when we need to know it. I can easily extend that to say We will only know what we need to know when we need to know it. In uncertain times we can never fully anticipate the situations we will face. This is an argument as to why humans evolved for abductive thinking: making unexpected connections between apparently unconnected things. The ability to do so is more resilient, and more adaptive, working to SOPs and the like. Think back to what I said about conceptual blending to reinforce this point.
In yesterday’s post, I advocated that capture of knowledge needed to be in fragmented observations and micro-narratives. That is also the best method of recall in a complex situation. Please note, that in the complicated and obvious domains of Cynefin you should be able to anticipate future needs and create conventional KM solutions. There are many tools and methods you can use there, in Cognitive Edge I have focused on the complex. So if you go back to yesterday’s examples of the use of SenseMaker® the medium of capture is also the medium of distribution. To find the knowledge you say something along the lines of If I knew the solution to this I would signify it in this way and that results in a recall of multiple observations from others that you can refine and conceptually blend to create a novel and resilient solution. Then the fitness landscapes allow you to connect broad patterns, but also (and critically) outlier events that may have significance.
You want access to real-world fragmented memories, not the synthesised retrospective coherence of best practice. The context of abstraction (in the Boisot sense of the world) allows for rapid codification and distribution but only within the context of its creation. So it works in complicated situations where the context is similar. But in the complex domain where the context is constantly shifting you need originally disinter mediated material, you want the original experience not someone’s interpretation of it. The good news is that if a recall is using the same signification structures as entry then the cost of a KM programme is (i) rapidly reduced and (ii) provides faster return and feedback. Critically it also means that innovation is a continuous process of discovery, not a centrally driven innovation initiative.
The trusted relationships that are part of those networks can only be treasured, social obligations built up over time through multiple projects are not something that can simply be recorded and they are not transactional anyway. I never liked the idea of their being a favour bank; the wrong metaphor all together. Many of them will only be known in the context of a need to know.
Trust is a complex aspect of human behaviour and again it is deeply contextual. I was taught Mountain Survival back at the age of 12 on the other side of Moel Siabod from the cottage in which I am writing this post. The instructor’s politics and attitude to women were appalling, so bad I can still remember it almost fifty years later! But in respect of navigation, shelter building and the like he was to be trusted. Trust can be relative to need, to the situation, to history. It takes years to win it, but it can be lost in minutes. We don’t trust people simply because the organisation has decided we should work together and be part of a community of practice.
I remember when I joined IBM I put a paper into the CoP on KM and it was rejected unless a more experienced person improved it in return for which his name would come first. I didn’t then realise that incentives were in place. I refused, the paper was not published within IBM but was within a KM magazine and was recently assessed as one of the top ten papers of all time in KM measured by citation. All of that without improvement by a corporate game player. Remove some type of content control and it becomes impossible to find things. Too many systems get initial take up but then get harder and harder to use as volume and participation increase. Does anyone remember the collaboration package Grove? That really suffered in that way and initiatives since by IBM, Google and Microsoft have all suffered from the same problem.
One study we did at IBM showed that people used the KM system to find the names of authors who they then phoned up to see if they should read the material. At a recent conference, I mentioned this and a KM advocate said that was no longer an issue as people only used the written material and didn’t feel the need to contact the authors. Aside from the fact that I don’t believe that, if it was true it would be dumbing down, and automation of knowledge which would be damaging. Human validation of codified material is needed. You need to trust the person, the office or the process which created the material. Those three are arranged in descending order of utility and increasing cost and time to create. An interesting alternative is self-discovery of real experiences as the process is your own, and it is something you have participated in. You want to hear people’s descriptions rather than third-party evaluations. The more material is summarised, the more it is interpreted, and the higher the need for trust.
Decision making under conditions of stress with limited information is always an issue and the authority or advocacy capability that comes with both experience and the lack of further ambition for promotion or power. This is frequently neglected in problem definition but it is important.
This one is critical, the more stress, the more I fall back on people to things that are familiar and have worked in the past. That may mean that new ideas get filtered or ignored. Now I could write a whole essay on this and will do over blog posts in the future so I can just touch the issue here at best. Go back to the links to MassSense and Crowdsensor to get a sense of what we can do here. Large-scale anonymous real-time feedback (which is not the same thing as social computing) means that I can use collective intelligence rather than trusted lieutenants to assist in decision-making. So we can use digital augmentation, but this also returns to the apprentice model. If I know that someone severed an apprenticeship, that they taught as a journeyman and that they executed their masterwork, then trust is easy. The more I lived the process, the more I understand it and its limitations. If I have served in the craft hall I also know the individuals and their capabilities. A world mediated by computers, formal roles and skills-based competences is an impoverishment of the wider knowledge capabilities of humanity.
Cognitive Edge Ltd. & Cognitive Edge Pte. trading as The Cynefin Company and The Cynefin Centre.
© COPYRIGHT 2023
So far in this series, I have been focused on knowledge in the context of ...
I’m feeling rather pleased with myself for getting back into the habit of daily blogging. ...
Leave a Reply