There's a 90% chance we won't panic
(appeared in the Times, 26th September 2011 - pdf here
Now that the rogue US satellite has crashed into the Pacific we can all come out from under our beds. The biggest bit of the satellite was about the weight of an adult gorilla, although not as soft, and travelled at 100 mph so it sounds rather ominous, but people only take up one 80,000th of the earth’s surface so it would be more than an unlucky day if anyone had been hit. 40 tons of debris got scattered over mainland USA after the Columbia shuttle disaster and nobody was injured, although NASA afterwards concluded there had been around a 1 in 4 chance of some casualties.
Stuff as big as this satellite plops down to earth most years without any publicity, although this time NASA’s policy of openness about the potential risks meant there was international interest. But can the public be trusted to react reasonably, or should they just be reassured? This is the basis of a crucial trial that started in Italy last week in which six top Italian scientists and a government official are accused of manslaughter following the earthquake in L’Aguila in 2009.
The issue focuses on a crucial meeting of the earthquake experts on March 31st 2009 - in the preceding weeks there had been many small shocks and a local amateur with home-made equipment had been predicting a major earthquake. The meeting concluded that ‘there is no reason to say that a sequence of small magnitude events can be considered a sure precursor of a strong event’, but at a press conference afterwards the official, Bernardo De Bernardinis, apparently translated this into the reassuring statements that there was ‘no danger’ and that the scientific community assured him it was a ‘favourable situation’. The succeeding events have the air of a scripted tragedy.
At 11 pm on April 5th 2009 there was a strong shock, and families had to decide whether to stay indoors or spend the night out in the town squares - the traditional response to tremors. Families who heeded the apparently ‘scientific’ reassurances remained indoors and 309 people were subsequently killed in their beds when the devastating earthquake struck at 3.30 am the next morning, flattening many modern blocks of flats.
The scientists are not being accused of failing to predict the earthquake, as it is acknowledged that this is currently impossible. The trial will instead focus on what was communicated to the public, which makes this all distressingly relevant to my job-title.
Who else might be culpable for issuing reassuring statements? Michael Fish jovially discounted the possibility of a hurricane in October 1987, and the subsequent storm killed 18 people.
In 1990 John Gummer was similarly reassuring about the safety of British beef when he fed his daughter Cordelia a beef-burger (although forensic analysis of the photographs suggests that the bitemarks in Cordelia’s burger are not those of a 4-year-old girl.) Over 100 people have subsequently died of variant CJD in the UK.
Of course there’s a difficult balance to be struck between reassurance and precaution. For every unheeded warning about a sub-prime crisis or cod depletion, there’s an exaggerated claim about the potential dangers of saccharin or the Millennium Bug.
And language is vital, and in particular the ‘framing’ of the risk. The L’Aguila risk was apparently around 100 times greater than normal for this town, but still only 2% chance for an imminent major earthquake. Similarly a modern heart operation has about a 2% mortality rate, but if a surgeon tells you there is a 98% survival rate it sounds much better. UK cardiac surgery uses survival rates, so that a surgeon with 96% survival still sounds good, whereas the US uses mortality rates, so the same surgeon would be reported with a 4% mortality, double the average! Similarly I might take an umbrella if I’m told there is a 20% chance of any rain, whereas ‘an 80% chance it won’t rain’ could lead me to leave it at home. To avoid these framing issues it’s recommended that both the chances of something bad happening, and of it not happening, should both be given.
So although the Italian prosecution seems harsh, it’s a salutary warning that people should be treated with respect, given full information and guidance for action rather than reassurances, and their concerns taken seriously.
When being trained to work on a phone helpline, I was taught the dangers of premature reassurance, since telling someone that it was all going to be OK before they had a chance to work through their anxieties closed down the conversation. Maybe scientists, as part of their media training, need to learn to listen as well as talk.