New Agora

For a while now, I have been thinking about the seeming conflict between the safe-fail experimentation approach advocated by Dave Snowden and Patrick Lambe's push for better accountability and professionalism in KM.

Both are talking about how KM can make a positive change to complex environments, but their emphasis on risk is almost 180 degrees apart. Dave's approach practically demands that people make mistakes to encourage discovery of positive changes. On the other hand, Patrick's talk identifies a complacency in KMers about the occurrence of mistakes that can have significant financial or personal impact, including deaths.

I said a seeming conflict because "safe-fail" experiments, by definition, mean experiments where consequences are controllable or reversible. So Patrick's high-stakes situations aren't places where safe-fail could be used.

What the two scenarios do illustrate is that different anticipatory responses need to be taken to prepare for mistakes, depending on the possible consequences.

Aside from obvious need to evaluate the potential scale of a problem that could be caused by a mistake, approaches will differ depending on whether the impact of the mistake is short-term or long-term. In the long term, some consequences may be reversible or at least not seen as so significant with the passage of time.

Bearing in mind that this matrix is not intended to be a comprehensive list of possible actions, some of the possible responses are:

Brief descriptions of each approach follow:

• Sandpit exploration - When consequences are minimal, people should be encouraged to explore and make their own mistakes in order to foster innovation.
• Safe-fail - Similar to sandpit exploration, but experiments need to be crafted in such a way that failure is toleration.
• Regulation - Most appropriate to indirectly control appropriate actions where long term impact will be serious if uncorrected.
• Embarrassment - A socially-enforced control used to impede inappropriate behavior that can have immediate impact (even if after time the impact will recede).
• Discipline - Behavior reinforcement techniques (negative or used) used to curtail mistake making.
• Injunction - Certain behaviors are forbidden outright, with the potential for severe personal repercussions if broken.
• Supervision - When immediate consequences are severe, an expert should be present to prevent or mitigate the impact of mistakes.
• Emergency override - Unlike Supervision, where people can still go against advice if they are sufficiently headstrong, the Emergency override takes control away from the mistake-maker to prevent the mistake happening.
• Redundancy - Where it is imperative that mistakes are prevented at all costs, redundancy is the best and only way to ensure that a single mistake has no unwanted consequences.