On the last day of the AAAS meeting, I sat in on a fascinating session on hazard prediction and risk mitigation. What’s fascinating, you ask, about risk mitigation? It sounds like something you read about on your insurance forms. But predicting danger—and particularly figuring out the best way to handle the uncertainty in those predictions—is at the heart of discussions over climate change adaptation. As Seth Stein said on the panel, “society plays a high-stakes game of chance against nature in a very uncertain world.”
Here’s a thing that I learned, not at this year’s AAAS, but many years ago at a psychology conference. The “right” trade-off for hurricane evacuation decisions, in terms of damage avoided and lives saved, is to order an evacuation when there’s a 20% chance of a direct hit from a strong storm. Twenty percent. That means that four out of five evacuations will result in everyone coming back and announcing, “Well, that was a lot of fuss over nothing.” And unfortunately, the one time that the evacuation really makes a difference, a lot of people will ignore it because of the four times it turned out to be unnecessary.
Getting back to AAAS, climate response is even harder than hurricane response because we don’t get multiple iterations. The full-on climate change version of Hurricane Katrina isn’t something we can learn from and do better with the next planet. (Okay, maybe it is, but the Hundred Year Starship seems like a hell of a plan B.) The one-off nature of climate tipping points means that we ought to be willing to act on small probabilities—and even more willing to act on the very large probabilities that we actually have for most of the really important climate hazards.
What’s really tricky is that we are still uncertain about some important aspects of climate change. This is very hard to talk about in public, because deniers will leap on any admission of uncertainty and treat it as uncertainty about climate change as a whole. This makes adaptation more difficult, because many of the real uncertainties take the form of, “We don’t know whether we’re at greater risk of Bad Outcome A or Bad Outcome B.” Other uncertainties… I was kind of freaked out, at an earlier panel, to learn that agricultural models only predict the effects of climate change for 10 or so of the world’s important crops—we have no idea about hundreds of others that we depend on for food and medicine.
Exploring those possibilities could help save lives—but only if we can learn how to treat uncertainty as something other than all-or-nothing.