“Don’t say this business about ‘No more tremors are expected.’ That’s a totally fucked-up thing to say,” he said. “You can never say anything like that when talking about earthquakes . . . not even under torture.”
Instead, he explained, he wanted a “media operation.” Just wait: Italy’s earthquake luminaries were on their way to L’Aquila. They would clean up the mess.
The meeting, attended by seven experts, including Selvaggi and Boschi and a handful of local officials, took just an hour and a half. Their conclusion: A major quake in the near term was unlikely. But remember, this is earthquake country: you never know. Boschi’s words during the meeting would later prove pivotal. “A large earthquake along the lines of the 1703 event is improbable in the short term,” he said, “but the possibility cannot definitively be excluded.”
An Abruzzo official pressed the prediction question once more. “We would like to know if we have to believe those people who go around creating alarm.” She was referring to the self-proclaimed expert, Giuliani. Such claims have no scientific basis, replied commission chair Franco Barberi. “The seismic sequence doesn’t foretell anything, but it surely refocuses attention on the seismogenic zone where, sooner or later, a large earthquake will occur.” The only thing you can do to protect people in such a place, he reminded her, is make sure structures are safe. As scientists and engineers repeat almost like a rosary: earthquakes don’t kill people; buildings kill people.
Residents of the town learned of the group’s assessment through news sound bites, including a snippet of an interview with commission member Bernardo De Bernardinis, whose background is in hydrology, not seismology. The interview was filmed before the meeting but broadcast afterward, giving the false impression that it was a summary. Was the swarm a sign of worse to come? “On the contrary,” De Bernardinis said. “The scientific community assures me that the situation is good because of the continuous discharge of energy.”
It was a wildly incorrect statement. Fault lines are not pressure valves, and small tremors don’t necessarily release energy that otherwise would have contributed to some bigger earthquake. Virtually all seismologists agree that there’s no known correlation—positive or negative—between the timing of small quakes and large ones. The television reporter, however, found De Bernardinis’s answer to be so satisfying that he decided to finish on a jovial note: “So we should go and have a nice glass of wine?” Oh yes, De Bernardinis said, recommending a vintage from his hometown.
Six days later L’Aquila and so many lives were destroyed.
Many of those who were in L’Aquila that night are haunted by the sound of the quake, an eerie and inescapable rumble that felt as if a supernatural force were bearing down on them. Others recall the red cloud that quickly covered the city, caused by countless red terracotta roof tiles shaking, falling, and breaking.
Once the shock had subsided, bodies were removed, and rubble cleared, survivors began speaking out against the commission. They insisted that it had been the reassurances of the experts that persuaded people to stay inside the night of April 5, even after the two tremors, rather than head outdoors and away from unsafe buildings.
One critic of the commission was Parisse, the journalist who had lost his teenage children. “As a father,” Parisse told me, “I’m the one ultimately tasked with keeping my children safe.” But that does not negate what he sees as a failure of responsibility on the part of the scientists and engineers. “They came to L’Aquila to reassure people, not to assess the risk,” Parisse says.
Many others felt the same. Soon a lawsuit was being put together by locals, accusing the scientists of negligence in their statutory duty to assess risk and advise how best to minimize loss of life and property. Parisse joined the suit. As a journalist he could write more articles, he says, but their impact would be minimal and ephemeral. “With the justice system, you have a chance for a more lasting truth.”
The following spring the seven men—five commission members and two experts, including Selvaggi—were charged with manslaughter. The claim: in the March 31 meeting and statements afterward, they had knowingly neglected their responsibility to inform the population about the risk at hand.
The announcement of formal charges triggered a flood of condemnation around the globe. “Risk of litigation will discourage scientists and officials from advising their government or even working in the field of seismology and seismic risk assessment,” declared an editorial in the American Geophysical Union journal, Eos. The CEO of the American Association for the Advancement of Science wrote to Italy’s president, reminding him that a veritable mountain of research, much of it conducted by Italians, shows that earthquakes are unpredictable.
But Fabio Picuti, Abruzzo’s 45-year-old public prosecutor, was either immune to the scorn or emboldened by it. Outsiders don’t know what happened here, he argues. They don’t know about Bertolaso’s order for a “media operation.” They can’t comprehend the full impact of De Bernardinis’s comment about “energy discharge.” They’re unfamiliar with our customs, our humility when it comes to Mother Nature’s destructive force.
As he put it: “The easiest media line is to say that this was ‘science on trial.’”
Abruzzo’s main courthouse was so damaged by the earthquake that the trial was held in a blocky set of temporary buildings in an industrial part of town. It was the summer of 2011. On a gate outside someone had made a sign: “To punish those who killed our children is not revenge. It is a way to let our children die a little less.”
Picuti’s case against the scientists built a pattern: residents resisted their established habit of fleeing their homes during tremors because of an overly calming message from the distinguished commission. It was “disastrous reassurance,” as Picuti likes to put it.
The trial was consumed by testimony from injured victims and the bereaved. People spoke of relatives who stashed blankets and cookies by the door to grab before exiting in the event of a tremor but had chosen to stay inside after seeing the TV interview. There was the man whose family had long since believed that tremors are followed by larger subterranean “replies”—and used the experts’ assessment to convince his pregnant wife there was no need to go outside that night. All of them, including their infant son, died when the couple’s home collapsed. There was the university student who was crushed to death, even though his friends had inquired about their dormitory’s seismic stability just a week before. Local officials had told them not to worry.
Assembling this testimony was essential to Picuti’s argument for manslaughter. He reasoned that the deaths were caused by things the scientists and engineers said and did not say. He cherry-picked past statements and snippets from scientific journals to argue that the scientists had known there would almost certainly be a major event. They couldn’t have known an exact date and time, of course, says Picuti. But they deliberately suppressed discussion of risk for the sake of reassurance.
During the 13-month trial, the prospect of being found guilty struck the men as preposterous. They didn’t dismiss the situation as unserious: people had died, after all. But the notion that they were guilty of negligence—of manslaughter—simply defied reason. “Nobody really believes that I would be so stupid as to go to Italy’s most seismically active region to say to the people, ‘Don’t worry,’” Boschi told me.
But Picuti deftly used the scientists’ own work against them. He showed the court a seismic hazard map of Italy produced by INGV, where Boschi and Selvaggi worked. Using a color-coded system from deep red (highest risk) to pale green (lowest), the map shows the probability of a major earthquake over the next 50 years.
At first glance two locations look downright dangerous: one near the southernmost part of the country, the other a wedge-shaped area that runs directly over L’Aquila. Picuti contrasted this level of risk portrayed by the map with Boschi’s “improbable” assessment.
When I asked Picuti about the map, its 50-year time horizon, and
the fact that day-to-day risk of a major earthquake remains low even in the areas colored deepest red, he was dismissive. “The people hear ‘low’ or ‘improbable,’ and then think: low.” He told the judge that what the public should have heard is what it says on the map: highest.
But Picuti’s masterstroke was wielding a 1995 academic paper called “Forecasting When Larger Crustal Earthquakes Are Likely to Occur in Italy in the Near Future.” Using historical records, geological evidence, and the best seismic data available at the time, seismologists tried to predict earthquakes for different areas of Italy over time scales of 5, 20, and 100 years. According to the model, the probability that L’Aquila would get hit with a major earthquake within all of those ranges was 1. That is not a typo. The model predicted a major earthquake in L’Aquila with 100 percent certainty.
The lead author of the paper? Enzo Boschi.
To hammer home this point, Picuti put another INGV seismologist on the stand. He summarized what Boschi and his colleagues had written but then surprised Picuti by explaining that the model was simply wrong.
A formula that produced a 100 percent chance of an earthquake occurring in the next five years would obviously give the same forecast for the next 20 or 100 years, he said. Yet no major seismic event had occurred during that initial five-year window. The quake’s nonarrival doesn’t mean that in year six it’s overdue and that in year seven it’s even more so. All it means is that the model itself is wrong. Boschi and his coauthors had even flagged their conclusion as suspect in the paper itself.
As Selvaggi watched this testimony unfold, he couldn’t help feeling hopeful. Picuti had trucked out evidence that earthquake prediction isn’t possible and then let a highly credentialed scientist deliver a lesson in probability and the scientific method to an audience that evidently needed one.
But Selvaggi was too optimistic. On October 22, 2012, Judge Marco Billi, an athletic 43-year-old with short-cropped black hair, walked to the front of the makeshift courtroom. Italy does not use juries: the decision was for Billi alone. Eyes down, he read his verdict in a barely audible monotone. For delivering “inexact, incomplete, and contradictory information,” the scientists and engineers were found guilty of involuntary manslaughter. They each received a six-year prison sentence, pending appeal.
As Billi saw it, the attendees of that 2009 meeting were responsible for the deaths. Not of all the victims—only those who Picuti could show had a habit of fleeing their houses when there was a tremor. The science underlying Boschi’s 1995 paper was of no interest to Billi. As he later told me, “We didn’t look at the details of the model. We only looked at what he [Boschi] wrote—that is, that there was a probability of 1 that L’Aquila will have a major earthquake. That’s all. It’s Boschi’s words!”
Boschi is furious over this mind-set. It’s not only that the earlier model was faulty, that Picuti can’t understand probabilities to save his life, or even Billi’s staggering contortion of logic. It’s that we shouldn’t be here in the first place, talking about research and scientific papers, the whole point of which is to share so that others can disprove or refine what you’ve come up with. “I am willing to go to jail for this point,” he thunders. “A scientist can write whatever opinions he wants in a scientific paper and it is off limits to a judge.”
Even in the land of Berlusconi and the judicial circus of cases like Amanda Knox’s, convicting a bunch of geoscientists in the wake of a natural disaster marks a new low. What would Galileo say? But what happened in L’Aquila is a window onto how we think about, communicate, and live with risk, and about impediments to clear thinking that afflict us all.
In the winter of 1951 a group of CIA analysts filed report NIE 29–51. Its aim: to examine whether the Soviets would invade Yugoslavia. And the bottom line? “Although it is impossible to determine which course the Kremlin is likely to adopt, we believe . . . that an attack on Yugoslavia in 1951 should be considered a serious possibility.” Once finalized, the report made its way into the bureaucratic machine.
A few days later a State Department official met up with the intelligence whiz whose team had composed the report. What did “serious possibility” mean? The CIA man, Sherman Kent, said he thought maybe there was a 65 percent chance of an invasion. But the question itself troubled him. He knew what “serious possibility” meant to him, but it clearly meant different things to different people. He decided to survey his colleagues.
The result was shocking. Some thought it meant there was an 80 percent chance of invasion; others interpreted the possibility as low as 20 percent.
Years later, Kent published an article in Studies in Intelligence that used the Yugoslavia report to illustrate the problem of ambiguity, particularly when talking about uncertainty. He even proposed a standardized approach to the language used for risk analysis—probable to indicate 75 percent confidence, give or take about 12 percent, probably not for 30 percent confidence, give or take about 10 percent, and so on.
Kent’s risk matrix never caught on, but the need for precise “words of estimative probability” is as relevant today as it was then. After the intelligence debacle that led to the Iraq War, for example, the U.S. Office of the Director of National Intelligence reworked Kent’s approach with new guidelines about the language of estimation. The goal was to minimize the blurring of information, probabilities, and confidence that had misled Americans into fretting about fictitious weapons of mass destruction.
Similarly, after Hurricane Sandy the National Oceanic and Atmospheric Administration looked at the consequences of its communications before the storm made landfall. Based on the National Weather Service’s classification system, Sandy went from being a hurricane to a post-tropical cyclone. That may be good for weather nerds and lexicographers, but for the media and the public the change simply served to confuse, especially when reporters mistakenly used the word downgraded to explain the change.
A May 2013 audit of NOAA’s performance emphasized that the agency should steer clear of abstract descriptions and instead make better use of language focusing on effects: possible flooding in these areas; storm surges up to this or that height; winds strong enough to down 100-year-old trees.
“If someone says ‘unlikely’ about an earthquake, you don’t really know what they’re talking about,” says Baruch Fischhoff, professor of social and decision sciences at Carnegie Mellon. Is it unlikely-like-having-twins unlikely, or unlikely-like-winning-the-lottery unlikely?
This fuzziness especially applies when talking about potential effects from natural disasters, because events like earthquakes and category 5 hurricanes are so infrequent. What will an earthquake of magnitude 6.2 mean for your life, your street, your kids’ school, your house’s foundation?
Conventional wisdom tells us that people are terrible with numbers. But as Kent realized back in the 1950s, we are even worse with words. In one study that Fischhoff coauthored, people had trouble understanding a 30 percent chance of rain. It wasn’t the probability that tripped them up but the word rain. Are we talking drizzle or downpour? All day or just part of the day? And over what area, exactly? (Communicating forecasts in Italian is extra challenging. In English we can use forecast instead of prediction to convey uncertainty. In Italian there is only previsione, which has a strong deterministic connotation.)
The divine cruelty of what happened in L’Aquila is that when Boschi said that a major earthquake was “improbable,” he was—and remains—correct. But where a career scientist hears the word improbable and knows that rare events do occur, a nonscientist hears improbable as shorthand for ain’t gonna happen.
Yet even the most carefully crafted communication from the Serious Risks Commission would likely have fallen short. Not because it would have failed to reach people or been met with suspicion, but because probabilities mess with our heads.
Even if we can comprehend a 30 percent chance of rain, or near-term odds like a coin flip, low-probability events are different. They have a “bewildering�
� effect on people, says Howard Rachlin, a professor emeritus of psychology at Stony Brook University. So we tend to lump them together; 1 in 10,000 sounds just as bewildering as 1 in 100,000. This is why people buy lottery tickets, even though the likelihood of winning is outrageously less likely than an event like a big earthquake in a seismically active region.
“All low chances seem the same,” Rachlin says.
When it comes to living our lives today or making plans for next weekend, behaving as if low probability is essentially zero chance isn’t necessarily a bad thing. We would be paralyzed otherwise.
But stretch that low probability over time—which is how earthquake risk is estimated—and confusion with low probabilities morphs into complete incomprehension. If you live in an earthquake-prone place for 10,000 days, the cumulative probability gets higher and higher, approaching 1 in 1. Our minds, unfortunately, have a hard time keeping up.
“We don’t see how these small things add up when you do them over and over again,” says Fischhoff. In study after study—looking at compound interest, unsafe sex, driving without a seatbelt, floods, earthquakes—we underestimate such cumulative effects. It’s one of those cognitive shortcomings calling out for a name. Maybe it should be called something like time-risk blindness.
How alert should we be to the influence of these blind spots? When the stakes are nonexistent—in a focus group meeting about how to market saucepans, say—there is no reason to be on guard for biases that lead our thinking astray. But when the stakes are high, says Fischhoff, like when communicating seismic risk, “we owe it to people to understand what the specific barriers are and how we can best get past them.”
This is where the scientists and engineers of the Serious Risks Commission went wrong, even if they didn’t realize it. They had no sense of how their words would land. They were used to closed-door meetings, and the commission’s mandate was to advise the Civil Protection Department, not the public. But once microphones and cameras were added into the mix, everything changed: they were now risk communicators, and whether they knew it or not, or what they might have felt about it, became irrelevant. (Unfortunately, says Fischhoff, another robust result in social science is that “people tend to exaggerate how well they communicate.”)
The Best American Science and Nature Writing 2015 Page 36
