QUESTION~How do you convince people to share failures?

There are a variety of techniques here. The key is not to ask people to share failure directly, that only works in a high trust environment and those are rare and deeply contextual. i.e. firefighters will share failure (Weick & Sutcliffe), but that is in the context of their work. You can only get a similar behaviour if you create the same context and few companies are prepared to burn the office down every morning at the request of the KM team.

In Cognitive Edge use a variety of techniques for this. They all work with SenseMaker™ but its not essential, on low volumes they are all appropriate. I'll pick on three to illustrate the point.

1. The use of archetypal character sets (derived from the culture) around which people tell stories. To take an example, instead of saying How did you fail on this project? That is a statement which is unlikely to get you more than a punch in the face. Instead you get people to tell stories about how the archetypes behaved in the situation. Most story telling traditions developed the archetypal form, the Mullah Nasrudin from the Sufi's is one of the best. If you do something stupid in Sufi society you don't say that you did it, you create a story about how Nasrudin did it. If its a good one it spreads fast and everyone learns.
2. Alternative history approaches in which project teams create timelines in which they failed (if the project was a success) or succeeded (if the project was a failure). It is important to realise here that the purpose of the exercise is to learn, not to extract the truth. Its a common mistake in lessons learnt programmes to focus on truth at the cost of meaning. In practice successful teams are unlikely to reflect what actually happened, their success will have filtered their reflection of the past and may result in key learning's being missed. By moving them into a fictional environment you increase the chance of picking this up. If you couple this with backwards story telling, used by the police and others to catch people out in lies, you can be remarkably effective without (and this is key) being overt. A common new age fluffy bunny facilitator technique is to want everyone to have the cathartic experience of confession as they reflect on failure and indulge in an orgy of my failure was bigger than yours competition. Such an approach is all about facilitator power games and has little to do with learning.
3. the latest approach, anthro-simulation was developed as an alternative to scenario planning. It places people in a simulated game environment in which we create multiple alternative futures. In these environments, participants draw on (and record) their knowledge of the past and its possible alternatives to handle environments in which they are doomed to fail. Although they don't realise it at the time. On each run the new scenario is constructed to be credible, but not one they planned for. This technique came out of anti-terrorism but is now being applied in commercial organisations (like all our methods it will be on the web site shortly and downloadable for free subject to a creative commons license). It can also be run in metaphorical environments but that would take too long to explain here

You don't. It hardly ever happens, and only in special circumstances. Executives who tell you they have created such an environment are delusional and starting to believe their own myths. I have know some cases, Lend Lease was one several years ago within its engineering community, but that was based on inspired leadership and a culture that had evolved over several years. Its not something that can be designed and it will always have limits.

The techniques I listed above are designed to work with reality, rather than try and create an ideal world. I always say when people ask (and they frequently do) "How do I create a knowledge sharing culture?" that they are asking the wrong question. You can't create such a thing in a world where ,at an organisational level, failure is punished and success rewarded. Within small trusted groups sharing happens naturally, and there are limits to such trust. Such groups are created accidentally as people work on projects, but there are complexity based approaches, such as Social Network Stimulation (SNS which can increase the density and occurrence of such groups in a more controlled way. This involves managing the channels for knowledge flow, rather than managing the knowledge itself. It is based on the assumption that if I can get everyone in the organisation to within three degrees of separation of everyone else based on having worked together then most KM problems will solve themselves.

The final point here, is that the use of narrative based techniques in the field under fire, capturing knowledge in fragments is more effective at capturing key learnings than in reflective environments such as lessons learnt documents and after action reviews. Here the process of reflection will distort what actually happened. This also allows you to capture learning in the forms of photographs and allegorical references to YouTube sites etc which can be remarkably powerful.