Black Swans and the Challenge of Mitigating the Unknown
I was originally going to write this article in the summer of 2012, after the Aurora shootings. I was gathering data for that piece when the Sandy Hook shootings occurred. I listened, in numbed sadness with the rest of the country while the pundits opined upon the reasons 20 children and six adults had their lives ended, suddenly, violently and without apparent reason. The more liberal media discussed gun control. Others stressed the mental state of the shooter. All were searching, if not for a reason, then a solution. Both are tragically, elusive. The answer may lie not in traditional risk assessment models but in a modified methodology that takes into account the unpredictable nature of such events, likely targets ( vulnerabilities) and application of appropriate mitigation measures.
The truth of the matter is that there is no facile way to solve the problem of random school shootings. Data is sketchy, trends inconclusive. Although such events seem to have proliferated in the last decade or so, the truth is that schools are safe. Despite a rise of 50 percent in people who reported being fearful of school violence, there was a 40 percent decrease on school-related deaths during the same period. Still another report pointed out that “during the school year 2008/2009 there were 38 school-associated violent deaths — in a population of about 55.6 million students in grades prekindergarten through 12.” According to the National Center of Educational Statistics, in fall 2012, more than 49.8 million students will attend public elementary and secondary schools. Of these, 35.1 million will be in pre-kindergarten through 8th grade and 14.8 million will be in grades 9 through 12. Statistically then, the ratio of deaths by school shootings to overall student population is extremely small. They are, in risk parlance, low probability events.
Risk is often expressed as the probability that an event will occur. More specifically, it is defined as the probability that a certain outcome will follow a certain event. The Department of Homeland Security uses the equation
R (risk) = T x V x C, where T = threat, V = vulnerability and C = consequence. Thus, risk is viewed as the product of three variables that are dependant on one another. Given the impossibility of eliminating risk entirely, the issue becomes one of deciding how much risk to accept, given budgetary constraints, probabilities and occasionally, political and social considerations.
As applied to school shootings however this model, as one study pointed out, has inherent limitations. Although it holds up reasonably well in predicting risk from natural hazards, “…defining the threat and estimating probabilities are inherently challenging because of the lack of experience with such events…” It is challenging to adequately define the threat of a school shooting incident. Data is scarce (i.e., there haven’t been enough of these events to produce reliable data, despite around-the-clock media focus when such events occur). Variables are endless. Methods, equipment and approaches to the target available to an attacker are almost unlimited. Threat, moreover, is sometimes characterized through an assessment of capability, intent, history and targeting, among others. Given the availability of guns, the muddled mental states of many of these shooters and a wide spectrum of opportunity, characterizing threat in the context of a school shooting incident is virtually impossible. What then, are we to do? We can’t protect everything, always. Also, more indirect approaches to the problem, such as behavioral analysis have limited applicability in a school setting.
It is arguably in the area of consequence where school shootings are most influential. Also referred to as impact, the after-effects of school shootings are undeniably dramatic and long term. The media, who bemoan the “evil” deed while interviewing 8-year old “eyewitnesses” , add immeasurably to the level of impact involved in these events. One article, recommending changes in our approach to school shooting incidents, makes the point that “…media coverage following school shootings can be extreme in both scope and intensity” .
The recent media frenzy at Sandy Hook underscores this issue: unrelenting press coverage of a deeply emotional and tragic event increases impact exponentially. A 24/7 news cycle and the intrinsic news value of the Sandy Hook shootings only added to the shock and sadness felt across the United States and the world. Thus, although school shootings are themselves intrinsically high-impact events, a ubiquitous press greatly exacerbates their consequences.
In sum then, school shootings are inherently high impact events, driven by threats that are often difficult to characterize, predict or mitigate. They are, in fact, Black Swans.
The Black Swan Event
Nassim Taleb, in his book The Black Swan (Second Edition), refers to events that are characterized by three attributes. First, the event is what Taleb refers to as an outlier, in that it “…lies outside the realm of regular expectations.” Next, such events carry with them extreme consequences. Finally, although improbable (or precisely because they are improbable) human nature demands that these events be predictable and explicated. As such, Black Swans are retrospectively explained and thereby considered predictable. But as Taleb goes on to note, they are not. “Black Swans being unpredictable (italics added), we need to adjust to their existence (rather than naively try to predict them)…There are so many things we can do if we focus on…what we do not know.” Dealing with a Black Swan problem is made easier, Taleb suggests, when we “…focus on robustness to errors rather than improving predictions.” “The bottom line:” he goes on to say, is to “…be prepared!...Be prepared for all relevant eventualities.”
The Sandy Hook Problem
If Taleb is correct, the “experts” will look back upon the Sandy Hook shootings and attempt to understand how such a horrific event occurred; what we missed that might have permitted us to somehow prevent, or at least mitigate the tragedy. The shooting will be “explained”, rendered knowable within the realm of predictive experience and added to other Black Swans that we have turned into the white ones with whom we’re more comfortable. Taleb refers to this habit as “the retrospective distortion”. “Police searching for motive” the headlines read this morning. They were referring to the Sandy Hook shootings, trying to ascertain why Adam Lanza, a 20-year old “quiet” and “withdrawn” youth would kill 20 children and six adults at a Connecticut elementary school.
This begs a question: How is this helpful? What does this information bring to the tragedy that might help mitigate other such incidents? Why do we continue to attempt to close loops with data which provide little, if any assistance to the issue of school shootings? And since we believe that events such as this demand some solution, what do we do? How do we mitigate risk where the threat is fundamentally unpredictable and impact is very high? In risk terminology, the low-probability, high-consequence event. Sandy Hook.
Asking the right question
One possible approach (there are no answers) might lie in concentrating on the areas where we do have some data; where there is some measure of predictability. If we examine vulnerability in the risk equation R = T x V x C, some useful information becomes available. If meaningful threat characterization is impossible (e.g., who will mount the next attack, where will it occur, what weapons will be used, etc.), and the impact will be high, (due to the nature of school shootings and the frenzied media coverage that inevitably ensues), perhaps the answer lies elsewhere. As Socrates pointed out, wisdom lies not in knowing answers but in asking the right questions. We may not know who, when or even why, but we can make an education guess as to where. Not “where” in terms of the next school, but where in terms of target. Regardless of who the next school shooter is, he/she will attack a school: college, elementary/high school, private school, etc. And the areas within the school will be similar; administrative offices, classrooms, the cafeteria/student center, libraries – wherever the targets (students, teachers and administrative staff) congregate. This being the case, perhaps hardening likely areas within schools would be the most productive mitigation strategy.
I teach a course in risk analysis for a small university. Every semester, my students conduct a risk analysis on the school itself. Sometimes the threat they are given involves a school shooting incident. In these cases, the risk mitigation measures their teams apply to the university produce very similar results. Some illustrative examples follow:
• Classrooms are constructed without windows or exit doors. In some schools, classrooms walls are glass.
• Classrooms have one door, which opens in and cannot be locked from the inside.
• Doors all contain glass which enables anyone on the outside to look (or shoot) into the classroom.
• Interior walls are constructed of plasterboard. A shot from a 9mm handgun (Lanza had two of these) would travel through most (if not all) of the building before being stopped.
• Outer school perimeters are seldom contiguous; nor are they usually monitored, either by staff or cameras.
• Day-to-day protocols, even if understood and disseminated appropriately are often not observed.
• Importantly, emergency measures with regard to school shooting incidents are seldom briefed down to those who may actually need to know them and are (even less often) practiced.
• Administrative procedures (supported by classes, lectures, etc.) designed to identify and act upon certain student behaviors might be implemented and carried forward.
• Security guards/police and even teacher volunteers might be armed and placed in the school.
• On a higher plane, appropriate legislation in response to legal issues. These might range from availability of weapons to information-sharing among health professionals.
• Especially in the case of elementary schools, students need to be taught to both recognize and communicate dangerous behavior to school authorities.
It is again important to note that these are illustrative examples, not meant to be considered all-inclusive or universally applied across the spectrum of school campuses – college, public or private - in the U.S. They provide a measure of what Nissan Taleb refers to as “prevention” and “robustness”. In their totality, addressing vulnerabilities such as the ones above provide a level of resiliency to school operations. Moreover, they are areas where data, albeit sparse, do exist. Attempts to “explain” school shooting events, even if successful (i.e., we discover what drove Adam Lanza to kill 27 people) are seldom instructive. Likewise, our ability to predict such events is limited, if it exists at all. What we are left with then, is a focus on those areas we know experientially or can deduce logically. These may involve the physical vulnerabilities of schools and classrooms, administrative and/or emergency procedures and targets. Applying mitigation measures to these “known” weaknesses results in a net gain; a hardening of the targets attractive to the school shooter. This approach is far more likely to achieve results than endless discussions about “motive” (the why) or personal characteristics (the who) or choice of location and weapons (the where and what) of people who perpetrate such atrocities. It is rather, a focus, as Nassim Taleb says, on the “relevant eventualities”.
School shootings, although occurring at an increasing rate, are uncommon and unpredictable. As such, they might be classified as “Black Swans”, events which are random yet produce extreme impacts. To the extent that this is so, traditional risk assessment methods are of limited usefulness. It is perhaps more productive to focus on what is knownabout such attacks (i.e., physical, administrative and procedural vulnerabilities, targets) and direct mitigation measures accordingly. Lacking a predictive capability, our focus should be on prevention, robustness and resiliency.
Bob Raffel is an associate professor in the homeland security program at Embry-Riddle Aeronautical University. For the last six years he's been teaching a course he developed on critical infrastructure protection and risk analysis.