In 2015, I researched and wrote about the risk to industrial society from geomagnetic storms—terrestrial phenomena that, despite their name, originate on the sun. Mild storms give us the ineffable beauty of the Northern Lights, and the Southern Lights too. Severe geomagnetic maelstroms might, some fear, knock out key satellites or cause continent-wide blackouts that would take months to undo.
I concluded that the risk had been exaggerated in the studies that gained the most attention. Still, given the potential stakes and the historical neglect of the problem, the issue deserved more attention from people in science, business, and government.
As I wrapped up my investigation, I tripped on a question of statistical method that I did not have time to fully explore. I have since put more time into the question. I just finalized a working paper about the results and submitted it to a journal. The upshot for the geomagnetic storm investigation is that I have modified my methods for extrapolating storm risks from the historical record. As a result, I have raised my best estimate of the chance of a really big storm, like the storied one of 1859, from 0.33% to 0.70% per decade. And I have expanded my 95% confidence interval for this estimate from 0.0–4.0% to 0.0–11.6% per decade.
More explanation follows.