Pitt’s bomb threat response based on beliefs, not science
[The Pitt News, Apr. 16, 2012]
If Pitt’s objective to hold safety as its “primary concern” actually meant ensuring safety before addressing other priorities, the University would close tomorrow. So would any other institution that applied the same prerequisite. That’s because no “commitment to safety” exercised by human beings with finite resources can actually produce “safety” in its purest sense. We live in an uncertain world characterized by constantly shifting, often unknown variables that all affect the probability of disaster; the only real escape from threats to life — and all of its wonders — is not to have it.
Given how uncertainty exerts particular influence on the current bomb threat saga, the many members of our community who describe the University’s response policies as interventions that “make us safer” must think again. Screening, notifying and evacuating might calm parents’ nerves and reflect courageously on our administrators, but the idea that anyone addressing this situation can wield an adequate grasp on the risk of a violent event, let alone the effects of policy on that risk, should be thrown out. And if we can comfortably accept safety for what it is — an unachievable ideal, the pursuit of which takes great sums from other things we value — we might be able to restore our University and aid in society’s broader struggle to meet the challenges of the Internet age.
The ability to determine whether an event will occur is indistinguishable from the ability to understand its workings. Across many disciplines, from psychology to stock market speculation, people attempt to understand the world by guessing at the factors they think underlie events, assigning weights (or probabilities) to each factor and seeing how well their “models” seem to explain real-life phenomena. Armed with models, we try to predict the future.
Such model-based prophesying works well when the objective is to learn broad characteristics of a system and when the factors are few and clearly observable. An example of “easy” prediction is your typical coin-toss scenario: “Heads you get a dollar, tails you lose a dollar; calculate your expected value after one day of flipping.” With the two factors and their weights self-evident (heads and tails, each with equal weights and opposite results), the answer is unequivocally zero dollars.
Once you start predicting narrow events that involve the behavior of free people, such as violent events on campus, the “expected values” take on a new, sobering level of uncertainty. Take the automotive violence that, unfortunately, every so often affects Fifth Avenue. Although we might be able to understand the general probabilities of getting hit by a car on Fifth Avenue, we cannot possibly produce a sure-to-be-accurate prediction for the number of collisions to occur in the next month. To find that, we need to intimately understand every last contributing factor, which could range from the clarity of road signs, to whether school’s in session to even the success of the Pittsburgh Pirates years before (some theorists would argue that every particle in the universe would actually have some say). To eliminate all uncertainty from your model, you’d have to possess infinite information about the nature of Fifth Avenue car-pedestrian collisions; that is, you’d have to be a sort of traffic god.
Uncertainty might plague all attempts at predicting events that involve human interaction, but it can be chillingly more virulent for certain attempts than others. The Fifth Avenue collision exercise might indeed be subject to uncertainty, but it’s a pretty mild case. Why? Because car-pedestrian collisions happen often enough for us to construct reasonably accurate models, thereby getting a “pretty good sense” (within an inevitable error margin) of the identity and weight of the contributing factors. To reduce road violence on Fifth Avenue, we can then target the “major factors” with policy, at least in theory.
But devising similarly educated policies to protect against a possible, devastatingly violent event at Pitt (like a bomb threat coming to fruition) is far more difficult — the current situation does not afford us an equivalent “pretty good sense.” And the last 50 years provided only two U.S. university bombings, neither of which was preceded by floods of emailed threats, for us to learn from. Without any higher education institution having ever experienced Pitt’s unique set of conditions, uncertainty is king.
At this moment, we cannot expect anyone — investigator, dean or chancellor — to identify the relevant factors or respective weights. With so little information regarding the factors underlying sensational campus violence, we also cannot confidently pinpoint the direction that the bomb threats or the University’s responses might have moved Pitt’s “safety level.” Accordingly, no Pitt response policy can be based on acceptably scientific analyses. Instead, all our leaders have to work with are under-educated guesses, in other words, beliefs.
There’s an unfortunate thing about fighting uncertainty with belief-based policies: For every belief there’s a counter-belief, often of equal reasonability. Some say the bomb threats indicate a will to do physical harm; others say the threats’ so-far-empty nature reveal perpetrators who merely seek to disrupt. Some say screening, notifying and evacuating has protected students; others say the structural and emotional disruptions ensuant from the policies encourage more threatening, potentially violent behavior. Essentially, whether we’re less safe now or whether Pitt’s policies “make us safer” depends on what you believe, not what you can prove.
The business of providing safety from devastating outlier events such as campus bombings is like playing chess blindfolded: To exist is to move pieces, and any attempt to gain real understanding of the happenings of the game or the impact of your moves comes at huge costs with only dubious benefits (which arrive as beliefs, not predictive models). If we try to control the game, we end up not making any moves at all.
Helen Keller, whose blindness and deafness might have rightfully prevented her from taking on an uncertain world, intrepidly discarded such fear-wallowing, writing, “Security is mostly a superstition...Life is either a daring adventure, or nothing.” As Pitt has demonstrated a willingness to jeopardize its educational values for the sake of an unconfirmable sense of safety whenever someone hits the send button on an anonymous threat, we can only hope Keller’s attitude will find future reception. Otherwise, we have enslaved our hopes and fears to the ever-present uncertain.
Write Matt Schaff at [email protected]. For a survey of how Pitt’s response policies allow the bomb threats to affect our community, read his April 10 column, “Repeated evacuations, notifications cultivate fear.”
If Pitt’s objective to hold safety as its “primary concern” actually meant ensuring safety before addressing other priorities, the University would close tomorrow. So would any other institution that applied the same prerequisite. That’s because no “commitment to safety” exercised by human beings with finite resources can actually produce “safety” in its purest sense. We live in an uncertain world characterized by constantly shifting, often unknown variables that all affect the probability of disaster; the only real escape from threats to life — and all of its wonders — is not to have it.
Given how uncertainty exerts particular influence on the current bomb threat saga, the many members of our community who describe the University’s response policies as interventions that “make us safer” must think again. Screening, notifying and evacuating might calm parents’ nerves and reflect courageously on our administrators, but the idea that anyone addressing this situation can wield an adequate grasp on the risk of a violent event, let alone the effects of policy on that risk, should be thrown out. And if we can comfortably accept safety for what it is — an unachievable ideal, the pursuit of which takes great sums from other things we value — we might be able to restore our University and aid in society’s broader struggle to meet the challenges of the Internet age.
The ability to determine whether an event will occur is indistinguishable from the ability to understand its workings. Across many disciplines, from psychology to stock market speculation, people attempt to understand the world by guessing at the factors they think underlie events, assigning weights (or probabilities) to each factor and seeing how well their “models” seem to explain real-life phenomena. Armed with models, we try to predict the future.
Such model-based prophesying works well when the objective is to learn broad characteristics of a system and when the factors are few and clearly observable. An example of “easy” prediction is your typical coin-toss scenario: “Heads you get a dollar, tails you lose a dollar; calculate your expected value after one day of flipping.” With the two factors and their weights self-evident (heads and tails, each with equal weights and opposite results), the answer is unequivocally zero dollars.
Once you start predicting narrow events that involve the behavior of free people, such as violent events on campus, the “expected values” take on a new, sobering level of uncertainty. Take the automotive violence that, unfortunately, every so often affects Fifth Avenue. Although we might be able to understand the general probabilities of getting hit by a car on Fifth Avenue, we cannot possibly produce a sure-to-be-accurate prediction for the number of collisions to occur in the next month. To find that, we need to intimately understand every last contributing factor, which could range from the clarity of road signs, to whether school’s in session to even the success of the Pittsburgh Pirates years before (some theorists would argue that every particle in the universe would actually have some say). To eliminate all uncertainty from your model, you’d have to possess infinite information about the nature of Fifth Avenue car-pedestrian collisions; that is, you’d have to be a sort of traffic god.
Uncertainty might plague all attempts at predicting events that involve human interaction, but it can be chillingly more virulent for certain attempts than others. The Fifth Avenue collision exercise might indeed be subject to uncertainty, but it’s a pretty mild case. Why? Because car-pedestrian collisions happen often enough for us to construct reasonably accurate models, thereby getting a “pretty good sense” (within an inevitable error margin) of the identity and weight of the contributing factors. To reduce road violence on Fifth Avenue, we can then target the “major factors” with policy, at least in theory.
But devising similarly educated policies to protect against a possible, devastatingly violent event at Pitt (like a bomb threat coming to fruition) is far more difficult — the current situation does not afford us an equivalent “pretty good sense.” And the last 50 years provided only two U.S. university bombings, neither of which was preceded by floods of emailed threats, for us to learn from. Without any higher education institution having ever experienced Pitt’s unique set of conditions, uncertainty is king.
At this moment, we cannot expect anyone — investigator, dean or chancellor — to identify the relevant factors or respective weights. With so little information regarding the factors underlying sensational campus violence, we also cannot confidently pinpoint the direction that the bomb threats or the University’s responses might have moved Pitt’s “safety level.” Accordingly, no Pitt response policy can be based on acceptably scientific analyses. Instead, all our leaders have to work with are under-educated guesses, in other words, beliefs.
There’s an unfortunate thing about fighting uncertainty with belief-based policies: For every belief there’s a counter-belief, often of equal reasonability. Some say the bomb threats indicate a will to do physical harm; others say the threats’ so-far-empty nature reveal perpetrators who merely seek to disrupt. Some say screening, notifying and evacuating has protected students; others say the structural and emotional disruptions ensuant from the policies encourage more threatening, potentially violent behavior. Essentially, whether we’re less safe now or whether Pitt’s policies “make us safer” depends on what you believe, not what you can prove.
The business of providing safety from devastating outlier events such as campus bombings is like playing chess blindfolded: To exist is to move pieces, and any attempt to gain real understanding of the happenings of the game or the impact of your moves comes at huge costs with only dubious benefits (which arrive as beliefs, not predictive models). If we try to control the game, we end up not making any moves at all.
Helen Keller, whose blindness and deafness might have rightfully prevented her from taking on an uncertain world, intrepidly discarded such fear-wallowing, writing, “Security is mostly a superstition...Life is either a daring adventure, or nothing.” As Pitt has demonstrated a willingness to jeopardize its educational values for the sake of an unconfirmable sense of safety whenever someone hits the send button on an anonymous threat, we can only hope Keller’s attitude will find future reception. Otherwise, we have enslaved our hopes and fears to the ever-present uncertain.
Write Matt Schaff at [email protected]. For a survey of how Pitt’s response policies allow the bomb threats to affect our community, read his April 10 column, “Repeated evacuations, notifications cultivate fear.”