Rolling the Dice

Two incidents from just before Christmas in 2009 illustrate the fact that companies are rolling the dice daily, betting worker lives against some fantastically ill-defined maximisation of returns.

In the first case, Canada Post demanded employees take down a set of Christmas lights from the top of a file cabinet, stating it was a safety issue. The associated press release stated safety of employees was of paramount importance, while the union rightly observed management needed to become more concerned about high critical hazards. While not obviously wagering lives here, the fact is they have, because they directed resources (including a chest-thumping press release!) toward this bizarre event. While argument is sound that Christmas lights present a possible hazardous condition, every penny spent on this nonsense is one less spent managing risks that will eventually kill someone. This misdirection of effort is an egregious affront against the very idea of management, and shows that Canada Post has no understanding of resource application for value return. They will have spent a thousand dollars on this, at least, draining valuable budget that could be applied effectively. That they congratulated themselves in a press release only further shows the low or absent management quality. A good manager would not ever allow such an issue to be treated with such false bravado.

In the second case, four workers in Ontario, Canada, fell to their deaths on Christmas Eve from a faulty swing stage. While it will be months before an inquest determines details, what was apparent early on was that the employees were improperly trained, the swing stage was improperly maintained, and basic safety equipment was not being used. One has to wonder what the company responsible for this travesty was hoping to achieve? Were they really that utterly ignorant of the risks to their workers, or were they convinced that the risks were acceptable? If the former they are criminally negligent as well as incompetent, and if the latter they are simply criminal. In this case they killed four people, wilfully, by way of ignorance, even if in the end it turns out the safety equipment was ignored by the workers. For some reason those workers had a cultural bias toward getting a job done, rather than doing it safely. The saddest aspect of this is that in killing four people, they devastated four families, which represents social costs that go beyond any compensatory penalties they will ever face. This type of incident is why criminal prosecution is possible in workplace injury and death cases, because nothing short of hard time would ever send a message about this kind of disregard for human life.

The lack of significant decreases in fatalities over the decades is indicative of how many companies are rolling the dice rather than stacking the odds in their favour. A large part of this is because safety, being treated as an activity, simply doesn’t return on the investment. You cannot prevent or suppress the impact of accidents when your entire model for doing so is focused on counting irrelevancies that have no provable relationship to the underlying causes of accidents. Theory states that the more proactive measures one does (meetings, inspections, etc.), then the larger the drop in the accident rate. Data, of course, shows no correlation, since there isn’t a connection between counting and changing conditions that increase risk:

Our research has shown doing more of the same inapt counting actually erodes worker trust in the safety program and its purpose, and consequently they begin to view all initiatives connected to “safety” as more of the same drivel. This, in turn, increases risks dramatically, because positive initiatives end up buried in noise. And when focus is lost by the workers, the systemic failures become commonplace. This “safety exhaustion” exists at so many levels in most organizations it is a large contributor to the decline in reasonable investment in actual practical operational risk-management.

What may be sadder than the static rates themselves is that there is no real mystery why these rates remain unchanged. It all comes down to failed safety programs, failing because of a combination of flawed assumptions about safety and flawed processes. The key issue is that safety is seen as a process rather than an outcome, meaning people are trying to manage the end-result of the process, rather than focusing on the process.

Amongst the flawed assumptions about safety is the extension of the mistaken focus on outcomes to provide metrics. Total Recordable Incident Rate (TRIR) exemplifies the problem, because it focuses on outcomes of failures as a measure of success. To put it in painful perspective, we have developed an entire industry around a statistic that rewards us for failing less, without ever asking why we are failing. It is similar to grading on a bell curve, and generating statements like, “Company A is the safest in the world this instant because they have only killed five people this year!” Shockingly, that kind of ridiculous metric is exactly what we use.

Another of the assumption failures is a little harder to come at in traditional safety programs because they simply have no comparative advantage to risk-management, because you cannot compare luck to management. Activity counting in traditional safety relies upon irrational aggregation without any relationship defined between the activities and the functional operational control measures. So, in essence, we reward ourselves for doing thirty extra inspections in a given quarter, without any way to answer the validity of the inspections. This lack of relationship between the underlying risks and their controls, and the activities, means we will never have a cause-effect clarity in traditional safety. In risk-management, though, the entire structure of the process is about defining and managing the relationships in a way that provides insight. In a risk-management model we still do inspections, but our inspections might point at an asset, which points at inherent risks, which point to known controls. So, every inspection (asset or generic) shows a chain of insight, and makes it possible to analyze the control measures. The assumption of a relationship is not the clarification of a relationship, and as long as traditional models focus on outcomes they avoid relationship definitions and accountability.

Extending from the false assumption that counts somehow imply management, is that reactive, traditional safety somehow implies management. Most traditional investigations, like their activity counterparts, have absolutely no control failure trace. And even when Root Cause Analysis is done, the problem is that there is still no direct relationship defined between those efforts and what must be managed to avoid and suppress risk. I can say that the direct root cause of an accident was inattention of the employee, but I cannot manage such a condition in and of itself; what I can manage, being the controls that might impart more employee focus, are not traced by the traditional safety reliance on form-filling. Risk-management, being more orange than apple, has an integral ability to do just that, while maintaining all the features of traditional investigation models.

Assumptions are dangerous because they fail the instant any aspect of the assumption is even moderately flawed. But even pretending that the assumptions are valid, and this pretence is enormous, the flawed process destroys the integrity of traditional safety programs. The focus on outcome, even to the point of using it to measure the efficacy of the process itself, creates a scenario where participants begin to behave against the interest of safety and in the interest of deflection. This leads to poor implementation at every juncture, because inability to prove value proposition one way or another makes it acceptable to create paper, regardless of its meaning; redirect focus in knee-jerk fashion to high visibility “controls” that are often ineffective, and given focus because they appear easy to achieve; generally ignore the quality of primary control processes, which are harder to do and have a longer value delivery; and to entirely ignore the directive management control mechanism.

We have not yet encountered a company that can actually prove whether high priority corrective actions are being done. Even worse, there are no corrective actions for high-consequence events, because there is a failure to recognize the consequences as focused on the outcome.

When you try to manage an outcome rather than a process, your efforts are entirely wasted, since an effect is not a cause. And contrary to what some might wish, rolling the dice is not a process, it is a risk. Choosing risk that is undefined, uncontrolled, and ultimately unacceptable is not management; management is about using defined process to ensure outcomes. The only thing rolling the dice has in common with management is that it ensures an outcome, which is the eventual failure conditions that will kill workers. All the lip service to the idea that there is no price on a human life flies out the window alongside the improperly harnesses employee when an enterprise approaches safety as something to be done, rather than something to be achieved.