Science in Mind

Should scientists be held accountable if their predictions are wrong?

Destruction caused by the 2009 L’Aquila earthquake at the Church of Santa Maria di Paganica
Destruction caused by the 2009 L’Aquila earthquake at the Church of Santa Maria di PaganicaCredit: Michele Nastasi, courtesy archivio Lotus international

Last fall, a former government official and six Italian scientists were found guilty of manslaughter in a trial centered on information they provided about earthquake risk just prior to the 2009 L’Aquila earthquake. The ruling, now being appealed, sent waves of alarm through the scientific community, sparking concerns that researchers around the world could be held accountable for giving governments advice about natural disasters, which are inherently unpredictable.

This Wednesday at the MIT Museum, a panel including an Italian architect, a seismologist from MIT, and a specialist in dispute resolution from Harvard Law School will discuss at an open public event what happened in Italy and what its ripple effects could be on planning, policy, and scientists’ advisory role to governments and planners.

Robert van der Hilst, head of the department of Earth, Atmospheric, and Planetary Sciences at MIT and a member of the panel, said the case attracted international attention because it could have a chilling effect on scientists’ communications with the public. If scientists feel they are really on the hook for predicting uncertain events, for example, they may bow out when asked to help policy makers and planners make informed decisions.

Advertisement—Continue Reading Below

“It’s very much in good faith. ... You try to do your best, working with the data you have, looking at all the uncertainties,” said van der Hilst. “It may discourage people from giving proper advice to people when asked. Expanding that, extrapolating further, it may even discourage people to go into those areas, and in the long term, it will have a really negative effect.”

There is currently no way to predict earthquakes, but seismologists are often asked the question.

In Italy, a scientific panel was asked what to make of tremors that occurred prior to the devastating L’Aquila temblor. The scientists did the right thing, van der Hilst says: They assessed the likelihood of a large, devastating earthquake, and found it extremely low. However, the message given to the public was deemed too reassuring by the Italian judge.

“The concern was not so much they did not predict the earthquake, but that they were rather cavalier in what they said,” van der Hilst said. For example, the BBC reported that a government official advised the public to go home and have a glass of wine, even specifying the type: “Absolutely a Montepulciano.”

It’s not easy to communicate uncertain risks to the public. For example, no scientist can guarantee that there is no possibility a massive earthquake will hit Boston this week. So should government officials send out a warning about a tiny, but non-zero risk of a catastrophic earthquake? It would not make sense for people to take protective actions and leave the area. And if every high-risk but extremely-low-frequency event were to be treated like a serious short-term risk, the credibility of governments and scientists would rapidly wither.

Van der Hilst said that scientists and professional societies are talking about what can be done to help scientists and policy makers communicate uncertain risks to the public accurately—without falsely reassuring them, and without inciting a panic. The obvious thing to do is to take long-term preventive steps: create buildings that will not fall down and kill people, for example. Another idea, he said, is to take a page from weather forecasting and issue regular seismic forecasts.

People are used to uncertainty with the weather forecast—they don’t sue the meteorologist if he or she is wrong about the high or the low temperature.

“People know it is uncertain; there’s a very high level of familiarity with that notion,” van der Hilst said. A seismicity forecast could “get people used to the background seismicity; there’s always activity going on.”

But the issue at the heart of the court battle over the L’Aquila earthquake also reverberates beyond the challenge of predicting earthquakes. Other complicated systems that evolve over time, such as the climate and the economy, present prediction challenges for scientists.

Learning to communicate both what the findings are in those areas—and their inherent uncertainties—might help protect scientists from landing in court if their predictions turn out to be wrong. That also might help insulate scientists from the wrath of the public, which too often assumes that if a prediction is wrong, science doesn’t work and is corrupt.

“TalkBack 360: Science on Trial” will be held on Wed., March 20, at the MIT Museum, 265 Massachusetts Ave. Cambridge, from 6 to 8 p.m.

Share