Dave Snowden Keynote: Big Data vs Human Data #KMWorld

Speaker: Dave Snowden, Founder & CSO, Cognitive Edge

[These are my notes from the KMWorld 2013 Conference. Since I’m publishing them as soon as possible after the end of a session, they may contain the occasional typographical or grammatical error. Please excuse those. To the extent I’ve made any editorial comments, I’ve shown those in brackets.]

Session Description: Will information come from the misty mountains of the internet or the cloud with no human engagement as Big Data suggests? Don’t we need human sensors to share knowledge? Our popular and provocative speaker discusses the cycles of techno-fetishism that try and ignore the importance of human intelligence, seeking to create the great algorithm which will answer the questions of life, the universe, and everything else. Big Data is important, but it’s only the start of the journey, and savvy organizations realize they need a synthesis of machine and human intelligence. Get lots of insights and ideas to take home to your organization.

NOTES:

• What’s the Role of KM? This talk is part of a series of talks Dave Snowden has given in an attempt to drag KM professionals away from fruitless activities, such as simply storing information. For him, the role of KM is to enable innovation and support better decision making by the organization. To do this, we need to manage the entire knowledge environment — not just the bits that are easily codified.
• Technology vs the Human: Technology in its place is good. What’s its place? To augment not replace human capabilities. As we use technology, it changes us. Consider the smartphone — it has become so deeply entrenched in our lives that it is almost an extension of our brains. And as we come to rely on it, we lose some capabilities we had before. In fact, human beings can lose specific capacities over the course of 1-2 generations. He cites the example of the slide rule. People trained to use it also developed the ability to understand numbers in a particular way. Further, it turns out that people who learn to use the slide rule seem better able to see errors in computer code. By contrast, people who use calculators exclusively do not develop either of these abilities. In this case, the slide rule augments human capability while the calculator replaces (and possibly diminishes) human capability.
• Re-discovering the value of the human brain: The human brain is 1.4 kg of fats and tissues. Yet it can still outperform many algorithms. The human brain has developed to do well at pattern recognition, not process information. Computers have been designed to process information. When we force people to absorb and store information, we are not allowing them to do what their brains are designed to do best.
• Narrative Learning: This is the best way of humans to make sense of the world around them. We tell stories to identify patterns and convey information. Some stories are oral, some are in text. However, don’t make the mistake of ignoring stories in drawings  — often they are the richest stories.
• Narrative Mapping: It understands the basic patterns by which humans operate and then helps identify which patterns could or should be changed by amplifying the useful and dampening the less useful. “To change a culture, tell more stories like this and fewer stories like that.”
• The Efficient Brain: There are a host of responses of the body and brain that happen without conscious thought. These are autonomic responses For example, if your hand gets near an intense heat source, you automatically move it. You don’t have to think about it first. This is an efficient way of operating because it helps the brain use less energy. (It is already a huge energy hog.) There are even more things that can become almost autonomous, with sufficient practice. For example, after 2-3 years, we can drive cars without much conscious thought. Similarly, it takes 2-3 years before we can reliably recognize errors in computer coding. This isn’t just an information processing issue. It is a matter of experience, training and judgment. This is why we need to bring back apprentice programs. They permit repeated practice and, most importantly, they create a reasonably safe environment in which to experiment and fail. This is critical because we learn more from failure than from success.
• The Brain Evolves: Our brains evolve to respond to inputs and the environment. For example, over 2-3 generations of constant input or practice, there are resulting biological changes in the brain that make that practice unconscious.
  • Aristotle: “Knowledge must be worked in the living texture of the mind, and this takes time.”
• Brain Constraints: The brain can handle only 3-4 concepts at one time. The only way to handle more concepts or more complex concepts is through aesthetics — through art, through metaphor. By abstracting things we can absorb much more complexity and nuance.
• The Impact of the Environment:  Place, our physical environment, can have a huge impact on who we are and how we work. As economics and the drive for cost savings are forcing people into cubicles or common work spaces, the new work environment can have the effect of eliminating diversity of thought.
  • Dissent is more important than consensus. It is a myth that everyone should be aligned. It is important to tolerate tension and support diversity of thought.
• The Problems with Current Approaches to Knowledge Management: We spend a lot of time trying to stop people from working in silos or encouraging them to share across silos. We should forget about it since we can’t stop them from working the way they do. The better approach is to have them create metadata. People tend to be much more willing to share metadata with people outside their silos. The shared metadata can spark new ideas on the part of the people who receive that metadata.
• The Power of Narrative: In Iraq, the troops had no use for doctrine. What they valued most was blogging from the front lines. The secret of narratives is that they can handle ambiguity, they can be complicated and messy. Because of this, you can convey more information and more complicated information. Further, each listener will extract from the story the elements that are most relevant to the listener in the moment. Years later, that person may not remember the details, but they will remember the gist of the story.
• Wisdom of Crowds: We have come to believe that crowd-sourced information is uniformly good. However, sometimes this so-called wisdom is nothing more that “the tyranny of the herd.” By way of example, consider the Dutch Tulip Mania,  South Sea Bubble,  and the recent sub-prime mortgage crisis.
• Human Sensor Networks: This involves using people to elicit oral histories from a larger group of people. In Wales, they are using school children to ask people in the community what matters in their community. This project will replace polling and focus groups. It will provide the basis for evidence-based policy-making. Better still, once this network has been designed and created, it can be reactivated later to provide answers to specific questions as the need arises. Further, these networks can be used to disseminate information rapidly.
• Proactive Foresight: Ideally, we ought to create networks that do more than provide restrospective coherence. We need to build networks that help us develop proactive foresight — the ability to sense what is likely to happen and then prepare for it.
• Repositories vs Networks: If you have a choice between building a repository or a network, choose the network.  Snowden: “Repository rhymes with suppository. Guess which is better?” On a more serious note, real-time data (gathered through the network) are more valuable than data that have been pruned and polished later. In fact, fragmentary data are hugely valuable, but they are often culled and lost forever in the polishing process.
• Big Data vs Human Narratives:
  • While big data can tell you what happened (e.g., Joe got on the subway at 8:45am), only stories can tell you why it happened.
  • Another problem arises from the way we tend to interpret data. Typically, we eliminate the outliers and look for the general trends. The problem with this approach is that the strategic opportunities and threats often exist in those outlying data points.
  • Search algorithms also disregard outlier data. They focus on the most commonly searched concepts and on popular links. What are we missing by disregarding the outliers?
• Exaptation: Adaptation is when we develop for a specific function. Exaptation is when we develop for a specific function and then that new capability is used for a completely different purpose. [Perhaps this is a human example of “off-label use”?] We need to create a KM ecosystem for managed exaptation.
• Judgment: We need to create trust and training to help people exercise human judgment. Human sensor networks allow us to express opinions on important issues before the political climate requires us to take a hard and fast position that has to be defended to the death.
• Focus on Designing an Ecology, Not a Machine: Think about people and computers working together in an environment, rather than building a system. If we fall into the pattern of letting computers do what humans ought to be doing, humans will lose the capability to do that which they must do. Respect technology, but respect human capability more. Design technology to augment human capability, not replace it.
• The final words go to Hugh McLeod:
  • “Change is not death. Fear of change is death.”
  • “What we Are is changing quickly. What we MUST BE, even more so.”