Not knowing we don’t know
To lack relevant knowledge without realizing it can have disastrous consequences. Consider, for example, the underestimation of the risk of a nuclear accident when constructing the Fukushima power plant in Japan.
By Digital Globe - Earthquake and Tsunami damage-Dai Ichi Power Plant, Japan, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=14630274
By IAEA Imagebank - https://www.flickr.com/photos/iaea_imagebank/8657963646/, CC BY-SA 2.0, https://commons.wikimedia.org/w/index.php?curid=56218685
The Fukushima Daiichi Nuclear Power Plant in Japan's Ōkuma, Fukushima Prefecture was sited on a high bluff over the ocean. The seawall around the reactors was 10 meters (33 feet) high.1
In siting the plant, the power company TEPCO and the Japanese nuclear regulators relied on a risk assessment that concluded that there was an infinitesimal chance of the region suffering an earthquake greater than magnitude 8.0, the strongest earthquake previously recorded. Further, the 10-meter seawall was deemed sufficient because there was no history of a tsunami exceeding that height.2
MORE POWERFUL AND HIGHER
Sadly, the Tohoku earthquake that hit the region on March 11, 2011, was magnitude 9.0, 10 times as powerful as the maximum earthquake that the power plant was built to withstand. And the resulting tsunami peaked at 13-14 meters (43-46 feet), 3-4 meters above the seawall.3
The active reactors were automatically shut down when the earthquake occurred. When the electricity supply failed, emergency diesel generators powered pumps to circulate coolant through the reactors' cores to remove decay heat.4
The resulting tsunami that hit 50 minutes later flooded the Units 1–4 reactor buildings and knocked out the emergency generators. The loss of coolant resulted in three nuclear meltdowns, three hydrogen explosions and the release of radioactive contamination. Radiation released to the atmosphere forced the government to evacuate 154,000 residents from a 20-kilometer radius around the plant.5
Large amounts of radioactive water were discharged into the Pacific Ocean. The long-term health effects of the disaster are as of yet unclear. In any case, decontaminating the affected areas and decommissioning the plant will take 30 to 40 years, according to TEPCO.6
TOO SMALL A SAMPLE
The nuclear plant had been built to withstand the strongest past earthquake, with TEPCO and the regulators not envisioning worse. In retrospect, the evidence relied on for siting the plant was insufficient. Evidence was missed that would have changed the plans for siting and protecting the plant: Subsequent research has shown that a wave high enough to overtake the reactors had hit the coast a millennium ago. Further, a probabilistic approach - asking what were the odds of it happening rather than had it ever happened in recorded history - would have dictated against siting the plant on the bluff.7
The Fukushima Daiichi Nuclear Power Plant disaster offers a stark example of what can go wrong when we rely on a small and biased sample of data.
We are prone to display insensitivity to sample size, that is, we expect too little variation based on what we see in small samples. We tend to use insufficient statistics, that is, we draw a statistical conclusion from a set of data that is clearly too small. We are trapped by biased generalizing, that is, we misassess the situation by drawing a conclusion from a biased or insufficient sample.
In short, a handful of historical observations of earthquakes and tsunamis are not sufficiently predictive of the nature of future earthquakes and tsunamis.
But why did engineers and regulators, supposed experts, fall prey to the trap of relying on insufficient and biased information?
SWAYED BY BELIEFS AND THEORIES
One explanation is that they, like all humans, are prone to making the fundamental cognitive error. We tend to underestimate the contribution of our beliefs and theories to observation and judgement. We often don't recognize when we've made an interpretation and that there are other ways that the information could have been interpreted.
The problem is that our past experiences color how we interpret evidence and new experiences. If we have never seen nor heard of an earthquake above 8.0 on the Richter scale and a tsunami higher than 10 meters, why should we expect these conditions in the future?
The belief that the worst had already occurred was disastrous.
NOTES
1 https://en.wikipedia.org/wiki/Fukushima_Daiichi_nuclear_disaster
2 https://www.nytimes.com/2012/10/13/world/asia/tepco-admits-failure-in-acknowledging-risks-at-nuclear-plant.html
3 https://www.livescience.com/39110-japan-2011-earthquake-tsunami-facts.html
4 https://www.ncbi.nlm.nih.gov/books/NBK253938/
5 https://en.wikipedia.org/wiki/Fukushima_Daiichi_nuclear_disaster_
(Unit_3_Reactor)
6 https://www.theguardian.com/environment/2019/sep/10/fukushima-japan-will-have-to-dump-radioactive-water-into-pacific-minister-says
7 https://www.nytimes.com/2012/03/10/opinion/fukushima-could-have-been-prevented.html
