19th century reproduction of a plate from Prodigiorum ac ostentorum chronicon (1557) by Conrad Lycosthenes
There is something fascinating about earthquakes: the sheer power and destructive force and the unpredictability. Coming from a country that is more prone to flooding by rivers and sea living in New Zealand is almost surreal at times. The second largest city Christchurch has had a few major earthquakes in the past year (one killed 181 people) and the day before Christmas there was yet another relatively big one followed by the inevitable aftershocks and more quakes; in the last 24 hours 41 quakes were recorded. The capital Wellington is sitting right on top of a fault line and is prone to quakes. By far the largest city, Auckland, where I live, doesn't experience major quakes, just minor ones, but the city is built on 50-odd dead volcanoes. Apparently, another one could pop up somewhere in the Auckland area at any time. The re-elected (26 Nov 2011) government, and the whole country for that matter, are hoping and waiting for the earthquakes in Christchurch to stop. Obviously, the Cantabrians want to get on with their lives, and the government is hoping that the reconstruction will provide a much needed boost of the economy. Unfortunately, reconstruction cannot and will not begin till all is quiet and insurance companies are willing to provide cover again. Problem is that nobody knows how long this might take.
People are starting to get fed-up and the frustrations are starting to show. For example, one columnist wrote in an article How long can this go on? that the GNS (GNS Science, a Crown Research Institute) should have warned people in Christchurch that another big one was coming. Apparently, the scientists at GNS had some information that suggested that this was to happen but they decided not to go public for at least two reasons: 1) they were not sure how strong the next quake would be and when & where it would happen; 2) they did want to cause unnecessary panic amongst the people that already were on tender hooks. Since earthquakes are notoriously hard to predict (e.g. see this article by Florin Diacu on BestThinking) this decision seems to make sense, even in hindsight. The level of evidence is important and this was not of the highest level by a long shot as pointed out in this blog On Levels of Evidence (by an Auckland General Practitioner).
When people lose their loved ones and livelihoods in an earthquake the emotions can easily flare up and turn into anger: why were certain buildings deemed to be safe? Why was certain land deemed to be safe? Why did the so-called experts not forewarn people of the imminent disaster? Some of these outbursts are now starting to come through here in New Zealand but the witch-hunt and search for a scapegoat were really obvious in Italy. As Michael Smithson wrote in his blog on BestThinking six Italian scientists are being prosecuted for manslaughter because they failed to warn of the big earthquake in L'Aquila in 2009 although the story is a little more complicated than that. The cold hard fact is that at present nothing and nobody can accurately predict earthquakes. The Boxing Day tsunami that killed over 200,000 people in countries in Southeast Asia took everybody by surprise; the large Japan earthquake and tsunami that led to numerous deaths, enormous devastation and a nuclear disaster was not forecast despite Japan being at the forefront of earthquake & tsunami prediction and warning.
However, there are always people that claim, usually in hindsight, to have predicted a natural disaster. This happens time and time again and it is interesting to have a closer look at these and similar claims of future events (usually calamities). We have all heard of the anecdotes of people refusing to board a plane because of a premonition that something bad was going to happen and that the plane indeed happened to crash. A possible explanation is that most people have such thoughts before going on a trip by plane and few actually 'miss' their plane (e.g. held up in traffic) and very rarely planes do crash but when it does happen the events can be falsely construed as 'linked' while in reality they were random events and purely coincidental. These coincidences only get noticed because of their close proximity (in time and space) to the real event. In other words, when somebody has a premonition that somewhere in the world at some time a plane may crash then this does not receive any attention. Similarly, the non-crashes (i.e. non-events) get no attention either; only the so-called positive predictions (in hindsight) get aired. Another example, which no doubt many people have thought of, is that when I dream the winning numbers of the next Lotto draw like this guy did. Suppose I never dream such thing and suppose I never buy a Lotto ticket but decide this time to give in (you never know your luck) and buy a ticket and win. Now, that would be freaky. If I had shared my numbers with my wife before the draw at least one person may believe me when I claim that I had predicted the lucky numbers. However, if I had shared these numbers with my colleagues at work, put them on Twitter and Facebook (neither of which I use) and hundreds of people had bought a ticket with the same numbers all that would have happened is that we would have had to share the prize money but it would still be equally freakish as before. Just because more than one or many people buy into or act upon my dream makes it no more of a coincidence than if I’d not shared it with anybody and kept it to myself. (NB the sharing or not sharing of my ‘dream' is somewhat analogous to another hypothetical situation "If a tree falls in a forest and no one is around to hear it, does it make a sound?") A Lotto draw is a purely random event, a stochastic process that cannot be predicted by any means, ever. However, if I have a dream that a volcano is about to erupt in Auckland on Monday and I am so convinced about this that I try to warn everybody about this few if any would pay attention. I might get a visit from the guys in white or from the men in black but most likely people would think I'm a nutter and rightly so. But suppose the volcano does erupt on Monday without any other warning. This is not a random stochastic event but the powers of nature at work. Is it possible that there are warning signs that some people and animals may be able to detect, perhaps subconsciously? Is it possible that scientists have got it terribly wrong and have ignored important facts and information?
I looked on Wikipedia to see what is known about earthquake prediction since I don't know the slightest about earthquakes. There are quite a few models that have been proposed over the years but without an experimental lab-based data it is extremely hard to build and validate these models as one can imagine. One thing that stood out, however, was that the statistical underpinning and validation of any predictive model is paramount. In other word, the level of evidence is important, which is along the same vein as the blog mentioned above. In New Zealand one guy became (in)famous because he claimed that he had predicted the Christchurch earthquake and was able to predict the coming (future) ones as well. His name is Ken Ring also known as the Moon Man because he's a self-described weather forecaster who bases his predictions on the lunar (moon) cycles. The problem with his theory, although it sounds reasonable enough, is that it has apparently been discredited (rejected) by scientific evaluation of past major earthquakes all over the world. It may play a minor role but that's as far as it goes. Mr Moon got vilified (e.g. see story 5 of nzherald's most read news stories of 2011 and in this sciblog, which also explains why so many people believed him and still do!) and tested the patience of hardened journalists on national TV. It resulted in many interesting responses in the media one of which is worth reading as it contains the opinions of several experts. One scientist in particular again reiterates that it is not as simple as taking one single prediction and linking it to an actual event. In other words, Mr Moon may or may not have accurately predicted the Christchurch earthquake but the validity of his predictions are only little more than my dreams about an erupting volcano in Auckland or the winning Lotto numbers. So, unless the prediction includes location, time, magnitude and probability of occurring it has no validity in the eyes of the experts (read: scientists). I think that's fair way of dealing with these predictions, particularly when once realises that we're not dealing with harmless Lotto numbers but major natural disasters that are potentially affecting thousands if not millions of people. In Wikipedia this is succinctly summarised:
If a plausible mechanism linking the observations with the predicted earthquake is not offered, the credibility of the prediction is diminished, but it may not necessarily be rejected. Evaluations of apparent successes must include a statistical estimate of the probability that the prediction came true by chance, which is often the case with predictions by amateurs. Whether a prediction is scientific or amateurish is not based on who makes the prediction, but based on how the prediction is made and tested. Predictions can be formulated either by defining the limits of the parameters probabilistically or by firm values.
In the field of cancer it happens from time to time that somebody claims to have found a cure. Quite often these claims are shrouded in secrecy and vagueness, supposedly to protect 'intellectual property', and the claimants get ostracised by the establishment. Generally, they have to find a safe haven in distant places, e.g. to avoid clashing with regulatory agencies (e.g. FDA), which to some extent adds to the mystery & mystique surrounding these hyped (miracle) cures. But desperate (and sometimes gullible) cancer sufferers flock to these places as if on a pilgrimage (which is perhaps not even that far from the truth). And lo and behold, apparently some patients do get cured. Since these centres of healing generally don't publish numbers that can be verified or tested for statistical significance it remains an unanswered question whether the patients are actually benefitting from the treatment or not. In addition, it is well known that spontaneous remissions do occur, without a clear medical or scientific explanation. Quite possibly the immune system has a role to play in these remissions. Fact is that they do occur so it is not inconceivable that alternative cancer treatments and spontaneous remissions coincide without a causative connection, which may have contributed to the anecdotal evidence of treatment efficacy. Other explanations that have been put forward are the fact that many of these patients have explored (and exhausted) many other conventional treatments and that some of the putative cures or improvements could be attributed to these previously received treatments (delayed effect). The point that I want to make is that, as with predicting earthquakes and other natural disasters (e.g. the end of the world), claiming a cure for cancer has to come with specific verifiable predictions that rise above the level of random prediction. In fact, this is the key of any scientific prediction: it should come with a probability that signifies how likely or unlikely the event/cure was to have happened by chance alone. Even the neutrinos travelling faster than the speed of light in vacuum have to pass this test or else one cannot be certain enough that the measurement result (i.e. deviation from the expected value) hadn't been caused by chance alone, i.e. that it was a false positive. If you’re interested in reading about this here is a nice blog, with comments. I would argue, referring to the above blog, that most of science operates at this level of evidence, i.e. at the level of statistical significance (or lack thereof).
Getting back to predicting earthquakes it seems that the lack of understanding, combined with the technical difficulties of obtaining the necessary data (i.e. many km’s deep into the earth), is a major limitation to building statistically robust models that can accurately predict earthquakes. Given the fact that many lives and many millions of dollars depend on improving these models I wonder why not millions of dollars are poured into addressing this unmet need (or maybe this is happening). Perhaps it is not a matter of money alone, something that cannot be fixed by paying huge bonuses. Perhaps it is not sexy or 'edgy' enough to start a career in earthquake prediction and to attract the best & brightest to this field of science. On the other hand, the weather forecast gets an awful lot of attention – on TV presented by good-looking (!) people (m/f) – but the models are extremely complex and based on probabilities, which require the calculating power of supercomputers. Clearly, seismology has a long way to go yet. In cancer the predictions (here it's called prognosis) are also predominantly based on the experience of the specialist rather than on hard model predictions with relatively narrow bands of deviation. In some ways it is better that way because it puts a human face to the very human issue of life and death and nobody likes to have one’s fate decided by the roll of a dice.