Inevitability of the Improbable

ResearchBlogging.orgDiscussion at the Mapping Unstable Ground workshop constantly came back to accurately identifying and communicating risk, and the difficulty of making practical policy decisions about risk mitigation. This tied in with thoughts inspired by In Terra Veritas pointing me at Hsü’s Catastrophic extinctions and the inevitability of the improbable [ref name=”Hsu”]Hsü, K. (1989). Catastrophic extinctions and the inevitability of the improbable Journal of the Geological Society, 146 (5), 749-754 DOI: 10.1144/gsjgs.146.5.0749[/ref]. The paper is a philosophy of science narrative, pointing out that even rare events happen given enough time, and that the age of the planet (or better yet, the universe) is a very, very long time.

The basic premise is that Intuition is Wrong. Intuition says that rare events don’t happen (The “Big One” won’t happen.), yet evidence resolutely points to those same events happening again and again and again over geologic timescales (6 megaquakes in Cascadia in the past 3500 years[ref name=”Atwater”]Atwater, Brian F.; Tuttle,Martitia P.; Schweig, Eugene S., Rubin, Charles M.; Yamaguchi, David K.; and Hemphill-Haley, Eileen (2003). Earthquake recurrence inferred from paleoseismology Developments in Quaternary Science, 1, 331-350 DOI: 10.1016/S1571-0866(03)01015-7[/ref] .) Instead of trusting intuition, a scientist needs to look at evidence, and appreciate that what seems like a rare, improbable event is actually an event that could plausibly and logically happen over the history of the Earth.

Hsü’s paper is a really fun read, and the argument is a good reminder for that probabilities get a bit odd when working on colossal timescales, but the paper was more focused on defining particular terminology for referring to events by probability (defining what exactly “likely” means in statistical terms) than in investigating the title’s claim of inevitable catastrophe. This is disappointing partly because the aptly-named On The Quantitative Definition of Risk[ref name=”Kaplan”]Kaplan, S., & Garrick, B. (1981). On The Quantitative Definition of Risk Risk Analysis, 1 (1), 11-27 DOI: 10.1111/j.1539-6924.1981.tb01350.x[/ref] is already a foundational source for sorting out risk assessment vocabulary. Kaplin & Garrick’s paper is higher-density, far more technical, and an absolute must-read for anyone working with risk assessments or struggling with understanding probability, but does not address the practical ramifications of working with geologic timescales. It is not a substitute for the paper I expected from Hsü’s title, and I’m left craving more inevitability.

I also found Hsü’s choice of using only impact extinction events as case studies less-than-thrilling. Yes, impacts kill everyone and everything, are incredibly rare, and have happened most certainly more than once in our very long history, but this same claim of low-frequency events happening eventually also applies to pretty much every other form of disaster. I would have loved the paper to take on the insane variety of energy build-up and release over time to investigate the inverse relationship between magnitude and frequency for disasters on different scales: impact events for incredibly the upper limit of totally devastating and thankfully rare events, stepping through megaquakes and tsunami, down to the downright common occurrence of floods and wildfires (that still have their own relatively rare, huge events, and tiny, common ones). This would emphasize the maxim that disasters happen everywhere all the time. All you can do is pick which ones you’re most comfortable with. (“…we are not able in life to avoid risk but only to choose between risks.[backref name=”Kaplan”]”) Disaster is inevitable, and risk relative. This is the paper I wanted to read.

Given that disaster is inevitable, we need meaningful ways of discussing the threat posed by different forms of hazard. Wilson & Crouch wrote a concise, high-density, and practical paper[ref name=”Wilson”]Wilson, R., & Crouch, E. (1987). Risk assessment and comparisons: an introduction Science, 236 (4799), 267-270 DOI: 10.1126/science.3563505[/ref] on expressing risk associated with various medical conditions (from electrocution and car crashes in the first paragraph through radiation, toxicology, and cancer in the next four pages). Although medicine is far outside my usual field, the examples required no technical knowledge to understand the concepts: “…if a man lies dying after a car accident, the risk of him dying of cancer drops to near zero.[backref name=”Wilson”]” This paper is a great introduction to the challenge of meaningfully comparing risks from different sources:

The purpose of risk assessment is to be useful in making decisions about the hazards causing risks, and so it is important to gain some perspective about the meaning of the magnitude of the risk.[backref name=”Wilson”]

Wildfires wreck havoc in California, British Columbia, and Australia every year, yet globally far more people die in floods. While GIS is creating all new advances in risk assessments for multiple hazards in a single location, that’s an idea for another day. Stepping outside of natural disasters, is reducing the number of deaths from car accidents more or less important than curing cancer? Which risk is more important to mitigate? How much risk is acceptable, and how does that change with the associated costs and benefits? This isn’t a question only for science, but requires integrating policy (something I’ll have more thoughts on after the 2011 Global Platform for Disaster Reduction) and societal values.

Related Reading

Certainty vs. Uncertainty: What “Supervolcano” teaches us about science and society by Erik Klemetti
Disaster Statistics by the United Nations International Strategy for Disaster Reduction
The Scales of Geoscience by Matt Hall at Agile*
Depth and Height by Randall Munroe at XKCD


[references /]

This entry was posted in Geoscience, Mathematics, Practice of Science and tagged , , , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *