I was lamenting to a retired Federal Aviation executive about the lack of risk-taking in the slow integration of drones into U.S. airspace. Replying, he told me a story that had remained with him for many years. In an earlier career, he was called to the scene of an accident with a model aircraft. It was a “control-line” model, meaning the aircraft flew in a circle of about 30 feet radius. The aircraft was both restrained and controlled by the two strings of the “control-line.” The line broke, and the aircraft departed on a tangent to the circle, fatally injuring a bystander. That incident drove home the inherent, and sometimes unanticipated, risks of unmanned flight forever in his mind.

Like many things, the utility of flight is balanced by the risk inherent in the operation. As a test wing commander in the United States Air Force, I knew this implicitly. The corporate safety review process emphasized a “two-axis” risk model in planning flight test programs. Those axes were probability of occurrence and consequence of occurrence. I departed from that model in my decisions then, and still do now. My model is single axis: consequence of occurrence. Why you might ask? I once had a flight test engineer brief me on the consequences of an in-flight failure of a fixed, external store on the aircraft. BUT…the probability of this particular failure was at 4 decimal places or 0.0001% chance of multiple events happening to lead to this critical failure. I asked: “Yes, but what is the procedure IF it happens?” Response: “It will never happen, sir.” My retort: “Come back with the procedure IF it happens.” He reluctantly did so, and on the first test flight following this safety board, my alert phone rang. Yes, the 0.0001% failure had happened. With the recovery procedure in place, return to base was uneventful for aircraft and crew.

The lesson for commercial industry in drone operations is clear: Do not rely on probabilities in assessing risk. I am not a statistician. Everything is 50-50 to me. Either it happens or it doesn’t. I rely on a singular axis: consequences of occurrence. Always plan for the eventuality to happen, no matter how slight the risk, and then minimize the danger of that eventuality. That is what my retired FAA executive had articulated so long ago.

And there lies the potential for negligence. The traditional calculus of negligence was fashioned by Judge Learned Hand in U.S. v. Carroll Towing, 159 F.2d 169 (2d Cir. 1947). There, he outlined a process for determining whether a legal duty of care has been breached. The formula is described by P x L > B, where P is the probability of loss, L is the cost of the loss, and B is the Burden (or cost) of precaution. A duty is created if the cost of loss times the probability of occurrence is greater than the cost of precaution. In the prior example, the probabilities were so low as to potentially eliminate any burden on my part to mitigate a potential liability. Using Judge Hand’s analysis, I would not have been negligent to accept the engineer’s assessment.

This simply does not work in aviation, manned or unmanned. While the mechanical probabilities inherent in a systems failure budget are definable, there are many aspects of flight that remain highly dynamic, rendering P only semi-definable. The real issue with the application of Hand’s Formula is the determination of L. Drone aviation loss calculus may range from a minor loss of a $50 vehicle to death, as my FAA friend noted. It depends on the characteristics of the operational environment. This drives P to a constant 1.0, and negligence is triggered anytime liability is greater that the burden of prevention.

If an unmanned vehicle flies only over sparse, unpopulated territory, L is defined only by the cost of the vehicle and its payload. Risk may be acceptable. If the unmanned vehicle flies over a crowd, L is directly related to injury mechanics. A 2 oz drone may fly with acceptable risk over people because L is inordinately low. The same is not true of a 100 lb drone over that same gathering. In that case, L is linked to significant injury or death, triggering the highest burden to prevent a negligent situation.

But the common negligence torts as applied to drones can also be largely mitigated. This is accomplished though comprehensive operational planning, technical investment and an appropriate risk management protocol, derived from experience and understanding of the flight environment and its inherent risks.