Predicting the Unpredictable: The Tumultuous Science of Earthquake Prediction Book

It would be valuable to have reliable warning hours, or even minutes, in advance of a major earthquake. If we knew a quake was coming, we could evacuate seismically unsafe buildings, de-pressurize gas pipelines, have emergency responders on duty, etc. Because of the obvious value, this has been a major research area for decades, drawing interest from professionals, amateurs, and cranks. Alas, nobody has been able to reliably predict earthquakes.

This book describes those efforts at prediction, how they were tested, and why the problem is hard. The author is a professional seismologist, in total command of her material, and does a good job sticking to the topic without turning the book into a general history of seismology or geology textbook. There is enough scientific background that even a very casual reader can follow the narrative, but not so much as to bore the knowledgeable. The prose is competent, but with a slightly excessive tendency towards cliche.

In the author's account, the frustrating thing about earthquake prediction is that there's lots of signals that might imaginably be useful (changes in the speed of sound in the crust, low-frequency electromagnetic radiation, leaks of geological gas, foreshocks etc), lots of anecdote that "this signal preceded this quake", but nothing that really pans out statistically. When earthquakes happen in places with very good data collection, no useful signals are detected. The best we can say is that if there is an earthquake, there is a 5% chance of a bigger quake in the same place within three days.

Part of what makes the narrative engaging is that it explores the fraught boundary between science and wishful thinking -- and how scientists exploit the ambiguity. Earthquake prediction is an exciting idea that governments are willing to fund it. Respectable seismologists have historically been willing to nod and wink and suggest that "while we can't promise anything, if you give us a lot of money we might accomplish something!" Likewise, while we can't predict quakes in the short term, we can make long-range claims. There will *eventually* be a major earthquake in Southern California, and geologists have find it challenging to express the right level of risk awareness. "There is a good chance of a major earthquake this decade" sounds like a clear prediction, but actually doesn't mean very much.

I found this a quick, illuminating, and enjoyable read. If you are interested in how scientists interact with the public on matters of obvious concern, this is a good book to pick up.