When talking to information about incidents and near misses, we must learn from normal operations so that we can remove risky situations long before they occur.
Learning from accidents and #incidents is a must to prevent these unwanted events from happening again. But let's also consider the following view (courtesy of Craig Marriott from his book Challenging the Safety Quo):
Roughly speaking, people work about 2,000 hours per year for a career spanning 50 years: 100,000 hours per career. A #accidentrate of one per million hours thus equates to about one accident per ten careers.
Such a rating is reasonably safe. I think most people would agree that working ten lifetimes before an accident occurs is acceptable. If this is acceptable, it means that by limiting ourselves to accident data only to learn, we have very, very little #information available to use. We can increase this by order of magnitude by investigating near misses as well, but it remains small. In any case, the events that cause fatalities often have no apparent near-miss precursors. However, learning from normal operations opens every hour of the day to useful information. There is so much more to learn, and it just takes a little more effort and commitment.
If we can learn from normal operations, we can capture and share #innovations while finding more risky situations and redesigning the task to take away the motive to perform them. By learning from everyday activities, we can get rid of these risks long before they happen.