The right and wrong way to think about climate risk
Aggregate risk is a difficult concept, but once it clicks for you it clicks
OK, my last post got some justified grumbles — justified and predictable. Of course, I realize that, like any extreme weather-related disaster, the LA fires have dozens of causes, and most of them are local.
But asking whether a given weather-related disaster was caused by climate change is the wrong way to think about it. Climate change doesn’t cause anything; it alters the aggregate risk of weather-related disasters. More often than not, it heightens the risk. (In far fewer cases —such as death from exposure to extreme cold— it can also lower the risk.)
To fixate on any one event is to miss the point: the same changed atmospheric chemistry that increases the risk of fire in California also boosts the risk of mangrove ecosystem collapse in the Mekong delta, and of coastal flooding in Osaka, and of crop failures in Sudan, and of thousands of others.
When any of those disasters happen, there’ll be somebody to argue the Vietnamese government’s mangrove-preservation strategy was flawed, or that Japan’s flood defenses were inadequate, or that Sudanese farmers were using the wrong kinds of seeds. And maybe they’ll be right, narrowly. But they’ll still be missing the point just as much as the myopic jerks talking about brush clearing in LA now.
The trick is that heightened risk has to be aggregated globally.
A hotter atmosphere may bump the risk of disaster A from 15% up to 18%, of disaster B from 1% up to 1.2%, and of disaster C from 87% to 92%. In each case, if that given disaster does materialize, it’ll give rise to a thousand be dumb fights about what exactly caused that one disaster, while climate change keeps quietly heightening risks of disaster everywhere.
I know aggregate risk is a tricky concept to wrap your mind around. Maybe a context switch will help.
If you drive a car for 10 miles sober, your chances of a serious accident may be 1 in a million. After three beers, maybe your chances of an accident are 100 in a million.
Three beers in, you’re bearing one hundred times more risk. And yet there’s still a very good chance you’ll make it to your destination alive and well.
If a million people do that same drive drunk, you’ll end up, on average, with 99 more serious accidents than you would have if none of them had been drinking. That’s why there are laws against driving drunk!
Ideally, we would like to apply the same sort of probabilistic reasoning to climate risk. Except we can’t, because quantifying climate risks turns out to be fiendishly hard. Weather is a chaotic system, inputs and outputs aren’t linearly related in the same way they are for something like drunk driving.
This is the crazy-making bit: the hard core of deep uncertainty that makes talking about climate risk so tricky.
Will a 2.5 degrees hotter-than-preindustrial-times atmosphere take the risk of mangrove dieback in Vietnam from 15% to 18%, or will it take it from 15% to 58%? Will it take the risk of catastrophic floods in Osaka from 1% to 1.2% or to 4%? Will it make the risks facing Sudanese millet farmers a little bit worse, or a lot worse? In what time frame? For how long?
Papers trying to answer those kinds of questions are published every single day. But weather is a chaotic system, so small changes in model parametrization give rise to hugely different answers. Different researchers using different methods can come to very different conclusions — not because one is right and the other one wrong, but because trying to pin down probability distributions in a chaotic system is a nightmare.
Climate policy has to be made under these conditions of deep uncertainty: experts can’t agree on how exactly inputs relate to outputs, let alone on the impact different policies will have on the probability distribution of outcomes.
I’ve argued climate innumeracy is one big reason our climate debate is fucked up beyond all repair. Deep uncertainty is the other. There isn’t a ‘correct’ way to think about tradeoffs between probabilities that can’t be quantified. It’s vibes all the way down.
Because of deep uncertainty, the best we can do is say that aggregate risk rises with the heat content of the atmosphere. By how much? Scientists and advocates can argue endlessly about that, but the truth is we really don’t know. Is a 2 degrees hotter world 25% riskier than a 1.5 degrees hotter world? Or 50% riskier? Or 500% riskier? That’s up to the rota fortunae.
We’ll learn the answer together.
There is, however, a possibility that a 2.5 degrees hotter-than-preindustrial world is much, much riskier than the 1.5 degrees hotter world we have now. Climate scientists disagree on how likely it is, but virtually all of them agree that it’s possible.
Extinction Rebellion types see this possibility as a reason to call for the immediate abolition of fossil fuel use. The call is wildly impractical, not to mention politically far-fetched. It is a cure almost certain to be worse than the disease: we don’t know how to feed eight billion mouths without fossil fuels. Still, while the remedy they agitate for is wrong-headed, a key part of eco-activists’ logic is sound: we need to take worst-case scenarios seriously, and we need to take action to prepare for those scenarios now.
That’s why I remain focused on the need to develop the scientific understanding and the engineering basis for slamming the brakes on global temperatures if the need arises. Because if 10 or 20 years from now it turns out we do need to lower the temperature quickly, and we haven’t developed the techniques to do so, we’re going to be very, very sorry.
"myopic jerks"? For somebody who theoretically wants to build the bipartisan political support necessary for your preferred policies to get government approval and funding in democracies, choosing to insult roughly half the electorate is a spectacular way to shoot yourself in the foot. Most especially since you aren't following your own advice here: taking action NOW to prepare for possible natural disasters in the future STARTS with the measures that are NOT "deeply uncertain", the measures we KNOW are affordable, effective, and politically sustainable: which incidentally, I've read the reporting of a climate modeling analysis showing that reducing the brush as suggested would lower SoCal's wildfire risk by more than twice as much as even a successful NetZero would. Bluntly, the local factors demonstrably vastly outweigh the global factors into mere statistical insignificance.
"We should start with the cheapest mitigations that will predictably provide the biggest reductions to our risks" is the same underlying argument for BOTH brush clearance against wildfires AND your proposed albedo salts & oceanic iron against climactic warming.
Seriously, the proper response to deep uncertainty isn't to obsess over what you can't know or control, but rather to get the maximum benefit from those things that you do know and can control. Similarly, the proper way to think about aggregate risk is to first remember that it is ultimately about expressing the predicted frequency of events. Aggregate risk does NOT answer "Will X event happen?" (causal), only "How OFTEN is it likely to happen?" (frequency).
The question "Will X event happen?" is relevant for planning mitigating measures. It tells a planner what the maximum event severity might be to require mitigation. The question "How OFTEN will X event happen" is also relevant, but mainly for how soon the mitigation needs to be ready and what the sustainment costs may be based on expected expenditures per event. Between the two, the latter is generally more important for effective planning because it rarely matters how many times the X event will happen if the mitigation isn't effective the first time. Having a fully funded and staffed fire department, working and stocked emergency water reserves, and sufficient defensive depth from brush clearance can be calculated to the standard of preparedness for the worst wildfire in SoCal history (plus a decent margin of error for reserve capacity). That's a causal mitigation: it works whether the next big burn is this year or a hundred years from now. It also works even if the frequency of such fires outright doubles or the severity increases into the reserve margin a bit. Proper preparation prevents piss poor performance.
Developing cost-effective geo-engineering solutions to climate change seems like a reasonable and important goal. I wish the climate change movement would focus on that goal instead of Netzero by 2050.
I am not convinced that that the risk that a 2.5 degrees hotter-than-preindustrial world is much, much riskier than the 1.5 degrees hotter world we have now, but it is always better to have a solid Plan B ready to go.
Same goes for asteroid impacts and other low-odds planetary hazards.