In the aviation industry, complacency kills. Accidents are so rare that most people in the aviation industry will never see a serious one in their careers. Aviation is so safe because of a culture of accountability and honesty. When problems appear, they are pointed out and the person who pointed them out is rewarded. Enormous piles of paperwork follow every plane around the planet, systems are rigorously tested and airlines compete for safety records. Other industries, even governments, are trying to copy this culture because it has been so successful.
Success means fewer accidents but fewer accidents can lead to complacency. When the relationship between an accident and its cause is not as clear as it used to be, it can be difficult to interpret available data to gain meaningful insights. On statistical models, aviation accidents look like blips, there are very rarely two accidents of the same kind in any given period. Using statistics to glean insights is difficult when the culture and safety standards of aviation are so completely different now than even 20 years ago. Incorporating old data into new statistical models invites a lot of confusion and statistical anomalies.
Finding effective ways to prevent accidents is therefore difficult. If predicting an accident is basically impossible because they are so rare, how do we find out what we should change to make things better? Sometimes, the response of aviation companies is a knee-jerk, sometimes implementing useless or needlessly expensive fixes for non-problems. This approach does not benefit us as much as it should.
The emerging trend in aviation safety is collaboration. Organizations, governments, and manufacturers are starting to share more information on safety features, experiences, and the routine challenges they face. This is a welcome development but it is not enough. There need to be further changes to the culture of aviation; the first should be more detailed investigation. As the two Boeing Air Max crashes recently demonstrated, a small problem in one program can have deadly effects that were not investigated with enough rigor to prevent a second crash. Our proposal is to apply the same investigative process and monitoring that goes into some of the more critical areas of aviation to the entire sector at every level.
Advances in machine learning, coupled with never before seen quantities of quality data, can produce insights into processes that would never have been noticed otherwise. To make this process effective, wide collaboration is necessary. This is only one part of the EU-research proposals of Safety Intelligence and Safety Wisdom. Safety Intelligence is processing all the sources of information an organization has to identify problems. Safety Wisdom is what to do with all that information, essentially it is the decision-making process that must come from the available data.
That word, data, is the cornerstone of advancements in aviation safety. It is now possible to monitor nearly all of the processes in the manufacture and maintenance of airplanes throughout their operating lives and beyond. Those data are useless unless there are enough of them to make comparisons and statistical analyses. To have enough data, we need to develop a culture that produces and maintains it properly and the incentives to collaborate made clear to every organization and individual in the industry. Only then can the insights into rare accidents be turned into an effective preventative culture.