October 23, 2012
Conviction of Italian seismologists – a nuanced warning
Posted by Austin Elliott
The big seismological news that started off this week was the guilty verdict in the trial of Italian seismologists and government officials indicted for multiple manslaughter after the M6.3 L’Aquila earthquake of April 2009.
Natural hazards scientists the world over greeted this news with shock and dismay, and science-savvy folks of all stripes have expressed outrage about the verdict, let alone the trial itself. Angry scientists commonly lament this “trial against science,” and bewildered editorialists decry the absurdity of convicting and imprisoning seismologists for failing to warn people of an impending earthquake, a feat which I’m confident people generally understand to be impossible.
The problem is… Much of this outrage grossly oversimplifies the case, and in fact is commonly based on a fallacious understanding of the trial. It behooves everyone editorializing upon it to understand the nuanced details of the situation. A news feature from the journal Nature last year illuminates these details, providing a complete narrative of the unfortunate incidents that led to this dismaying conviction.
Scientists on Trial: At Fault? – Nature News Feature
In fact the trial was not about failure to predict an earthquake. Rather, the scientists were prosecuted for not fulfilling their alleged duty to properly prepare the populace for an earthquake. This is not so subtle a distinction. This reality shows the case to have more merit than the straw-man accusation of “failure to predict an earthquake,” but it still raises some important questions about how scientists should be expected to act when communicating to the public.
A fantastic synopsis of the incident, the trial, and its implications for the future of expert-to-public communication is publicly available to listen to from the Seismological Society of America, and is an essential prerequisite for informed discussion about this trail.
In that recording, Tom Jordan, director of the Southern California Earthquake Center and chair of the international group commissioned by the Italian government to report on the status of earthquake forecasting operations (ICEF: the International Commission on Earthquake Forecasting for Civil Protection), presents the short history of this incident and discusses the sophisticated questions it raises. His commission’s report is also available to view and is a more technical document, but an important read if your profession involves natural hazard or risk assessment.
Operational Earthquake Forecasting – State of Knowledge and Guidelines for Utilization – ICEF report
Okay Then, What Happened?
I highly encourage you to read the Nature article or listen to Tom Jordan’s SSA talk, but if you can’t muster the patience for those, this synopsis of those synopses may serve as a rough outline of the incident.
L’Aquila, a city in one of Italy’s most earthquake-prone regions, had a series of tiny earthquakes beginning in October 2008. The series escalated through March 2009, during which time a local man started issuing earthquake “predictions” based on radon measurements. These unfounded warnings had no scientific merit but began to alarm the populace as the region continued to suffer frequent small tremors. On March 30, fearing an unwarranted panic, civil protection officials cited this man for inciting public alarm and “forbade him from making any public pronouncements.” In an untimely misfortune, a 4.1 earthquake then struck the region.
Fearing further alarm, the head of the Department of Civil Protection called for an immediate meeting of the National Commission for Forecasting and Predicting Great Risks, in L’Aquila. Given the fluctuating spread and suppression of misinformation, the lack of accurate information, and thus the context of alarm and confusion, it’s not surprising that the press’s involvement at and around this hastily arranged meeting spelled doom for reasoned communication of the real risks. As Tom Jordan et al.’s report mildly explains,
“[during an earthquake sequence] earthquake probabilities may vary over orders of magnitude but still remain low in an absolute sense (<1% per day). Translating such low-probability forecasts into effective decision-making is a difficult challenge.”
Indeed the seismologists and civil protection authorities apparently grappled with how to explain that this ongoing earthquake swarm did not substantially increase the already substantial regional hazard of earthquakes. They discussed the empirical history of earthquake swarms and how they rarely (<2% of the time) resulted in large mainshocks, but were careful to state that that specific risk couldn’t be excluded. They turned their focus instead to dispelling the alarm caused by untested and poorly vetted prediction schemes. Unfortunately, they let the audience guide their discussion. When faced with the unanswerable question of whether there was an imminent large earthquake to worry about, phrased with disarming cultural charm–“so we should go have a nice glass of wine?”–the vice-director of the Department of Civil Protection was jovially lured into a fatal appeasement of the citizens’ fears. “Absolutely.”
Then the most unfortunate coincidence of all occurred. This was one of those <2% when the swarm unleashes a big one. That night (April 6) the M6.3 roared out from the hillsides and collapsed 20,000 buildings in and around L’Aquila, killing 309 people.
The prosecution was brought by several men who had lost their entire families after deciding notto sleep outside as would have been customary after the small preceding shocks. As one of them claimed,
“[the messages from the commission meeting] may have in some way deprived us of the fear of earthquakes. The science, on this occasion, was dramatically superficial, and it betrayed the culture of prudence and good sense that our parents taught us on the basis of experience and of the wisdom of the previous generations.”
Of course it is absurd to lay this loss on the shoulders of scientists, when so many individual decisions and uncontrollable instances led to these specific deaths. A more direct culprit that comes immediately to mind is earthquake-vulnerable construction. The twist here is that the populace had already accounted for that in their cultural custom of sleeping outside after quakes. In their eyes the “official word”–twisted and accidental as it had been–contradicted their understanding of safety “procedures” and misled them into harm. In essence, the scientists failed to fill an information vacuum: they dispelled the alarmism but didn’t replace it with the facts, or even appropriate guidance about the existing earthquake risk. Whether it was their civil duty to do so, and whether their failure to constituted criminal negligence or worse, was subsequently left up to the courts, and now we see how that turned out.
The ICEF report explains the problem precisely:
“Information vacuums can spawn informal predictions and misinformation, and … relying solely on informal communications between scientists and the public invites confusion. In this context, the deployment of systematic and transparent procedures for operational earthquake forecasting must be seriously considered.”
The series of unfortunate incidents and hastily orchestrated public outreach that led to this dismaying conviction invite challenging questions about the role of scientists in public policy and the complicated endeavor that is risk communication and mitigation. They also underscore the need for unified (or uniform) and culturally aware procedures, and offer some lessons that will guide the development of these. Understanding what to tell people about inherently unpredictable risks is hard enough, but when colloquial notions about the risk supplant proper information from authoritative sources, the reversal of these culturally ingrained practices becomes truly formidable.
At the end of the ICEF report, the authors lay out a roadmap for responsible operational earthquake forecasting. Their guidelines are informed by current technologies and practices, and call for continued efforts to test quake forecasting models. In the end they leave us with a sort of mantra of the seismological community:
While the responsible scientific research on earthquake predictability should be encouraged and operational forecasting capabilities should be developed, these activities cannot substitute for civil protection actions well in advance of earthquakes, for example in the design and planning of new buildings, or retrofitting of older ones identified as being at risk. Preparing properly for earthquakes means being always ready for the unexpected, which is a long-term proposition.
With a full understanding of the L’Aquila debacle, I expect that seismologists and the public can maintain a productive and orderly discussion as we develop more useful ways of understanding and communicating earthquake hazard. California’s got a big head start, since we’ve been training you for decades what it means to live in Earthquake Country.
[…] Austin Elliot (The Trembling Earth): Conviction of Italian seismologists – a nuanced warning […]
Excellent article that explains in plain English what happened, points to last year’s Nature news feature, and emphasizes what needs to be done. In addition to research on earthquake predictability, the development of operational forecasting capabilities, and civil protection actions there is an important role for widespread education of people who live on/near plate boundaries. Even though earthquake scientists have learned a lot about earthquakes we have witnessed in the last few years an enormous number of deaths, injuries and destruction caused by earthquakes–Indonesian earthquake of 2004 and Japan earthquake of 2011. It is clear that we have a lot of work to do to communicate what we know and don’t know about earthquakes to those who live on/near plate boundaries and to those responsible for protecting them.
Thanks for this post, since almost nobody in Italy is telling the truth about this trial and the reason is simple: almost nobody read the trial documents, which were available, indeed.
Here’s the prosecutor’s memoria: plain, clear and not lacking of scientific approach, for sure.
“In fact the trial was not about failure to predict an earthquake. Rather, the scientists were prosecuted for not fulfilling their alleged duty to properly prepare the populace for an earthquake. This is not so subtle a distinction.”
All due respect, there is no distinction if an earthquake is not predictable. One should tell people to go about their lives as they did a year ago as they did 10 years before that.
Taking temporary emergency measures only makes sense if you are dealing with a predictable phenomenon. Temporary entails predictive ability.
But there seems to be some oscillation in your article with regard to the predictability factor. You write:
“Earthquake swarms resulted in large mainshocks <2% of the time."
Roughly 2% of the time is not an insignificant number. You are stating that there was in fact an increased risk of an earthquake. You are stating that there is in fact a predictability factor with regard to earthquakes. A reasonable person might take some temporary emergency precautions if there is a scientific bases to raise the risk of an earthquake to that level.
Which is it? Was there an increased risk of an earthquake or not?
I think the ICEF report says it best: “[during an earthquake sequence] earthquake probabilities may vary over orders of magnitude but still remain low in an absolute sense (<1% per day). Translating such low-probability forecasts into effective decision-making is a difficult challenge.” In the scheme of precautionary measures that individuals can take, why would individuals change their behavior if the probability of an event changed from 1 to 2%? Considering the precautionary measures one can take, this is not a significant change. The scientists failed to convey that there was any risk at all, which there always is.
You're exactly right that they had no reason to do anything differently than they did one year or 10 years before… except that the quake scientists had not only an opportunity to convey the well understood earthquake risk, but they in fact both lost that opportunity AND went a few steps in the other direction, inadvertently communicating to the public that the risk was actually diminished.
If I expected to be killed about 1 out of every hundred times I walked out my back door, I would stop walking out my back door. So “1 to 2%” is quite significant for deadly events, you can’t just brush it aside.
Your reply here creates yet another point of confusion. Is the “normal” probability of a serious earthquake already 1%? From your previous statements it seems that the answer is no, hence why people sleep indoors except when the earthquake risk rises by several orders of magnitude. Just like the scientists, you’ve once again clouded the issue by trying to give simplifications instead of providing specific numbers.
Scientists most certainly do give specific numbers. The extent to which these small numbers are useful in guiding behavior is a very different question. The probabilities of earthquakes (of experiencing various levels of ground shaking, actually) over given time periods are available from government agencies that review and synthesize enormous volumes of data, and vary hugely depending on where you’re located and how well studied your region is.
Hazards in Italy can be found here: http://www.mi.ingv.it/pericolosita-sismica/
For the U.S., look here: http://earthquake.usgs.gov/hazards/
As you’ll quickly find perusing those sites, shaking probability estimates and calculations are highly technical, and generally intended for engineering purposes. These maps and the various accelerations and probabilities of exceedence plotted on them may seem additionally obfuscating, but that illustrates the extremely complex nature of seismic hazard, and, frankly, reflects the impressive amount we DO know about earthquake recurrence, mapping in great geographic detail subtle differences in the probability of earthquakes.
I can’t give you a simple answer about “the ‘normal’ probability of an earthquake” without a whole host of other information, and frankly, I think my point still stands: even with a precise report that the seismic hazard at your house today has increased from 0.01% to 0.5%, would you change your behavior?
Let’s take your analogy: If you expected to be killed once out of every 1000 times you walked out your back door, would you stop walking outside? Based on your response above, I suspect you would. So if I then told you that your risk had increased to once out of every 100 times, would that change your behavior? No. You’d still avoid walking outside.
This ill-fitting analogy warrants a clarification: earthquake probabilities do not report your risk of death, which is EXTREMELY small when you consider the death toll of earthquakes compared with the number of people around the world who have experienced deadly earthquakes. So we’re not talking about your risk of being killed upon walking out the door. It’s the risk of, let’s say, an anvil falling on your doorstep. If you do nothing about it, it will kill you, but only when it happens that one-out-of-100 time. Since you have to leave the house, and you don’t know which of the 100 or 1000 or 1,000,000 times you walk out the door the anvil will fall, you side-step it every time. Now let’s translate this back to Italy: you know there will be earthquakes–this is one of the most quake-prone regions in the world–but you don’t know when. It could be one every ten years or one every 1000 years, but you don’t know which year it will be, let alone which day. You have to earn a livelihood by going to work daily. You have to shelter yourself in a house nightly… you can’t be paralyzed in an empty field because at some point there will be an earthquake, and for the purpose of planning your daily life, the probability of an earthquake at any usefully predictable second is still slim to nil.
There’s a new paper (2015) that discusses more practical approaches to dealing with and communicating low probability events. It deals specifically with the issue of probabilistic hazard versus individual risk of death. http://srl.geoscienceworld.org/content/86/6/1674.extract
Unfortunately, most of it is behind a paywall.
Great article by Mark Quigley, an earthquake scientist in New Zealand, that addresses this very issue, informed by his exceptionally relevant experience with it. http://theconversation.edu.au/laquila-charges-leave-earthquake-scientists-on-shaky-ground-10301
[…] Some of the complex undercurrents surrounding the L’Aquila verdict are nicely summarised at The Trembling Earth blog. http://tremblingearth.wordpress.com/2012/10/23/conviction-of-italian-seismologists-a-nuanced-warning… […]
[…] likes of social scientist Brigitte Nerlich, statistician David Spiegelhalter, and seismologist Austin Elliot (the latter’s analysis is particularly […]
[…] 고수한다는 것은 사실과 다릅니다. 사회학자 브리짓 넬리치와 지진학자 오스틴 엘리엇의 글은 이번 사태에 대해 의미있는 분석을 하고 […]
[…] news, Italian officials are now trigger-happy with evacuation orders in the wake of the manslaughter conviction of seismic hazard officials. Caution is good… but, this is why we have legends like the boy who cried […]