In Reasons and Persons, philosopher Derek Parfit wrote:
I believe that if we destroy mankind, as we now can, this outcome will be much worse than most people think. Compare three outcomes:
2. A nuclear war that kills 99% of the world’s existing population.
3. A nuclear war that kills 100%
2 would be worse than 1, and 3 would be worse than 2. Which is the greater of these two differences? Most people believe that the greater difference is between 1 and 2. I believe that the difference between 2 and 3 is very much greater… If we do not destroy mankind, these thousand years may be only a tiny fraction of the whole of civilized human history.
The ethical questions raised by the example have been much discussed, but almost nothing has been written on the empirical question: given nuclear war, how likely is scenario 3?
The most obvious path from nuclear war to human extinction is nuclear winter: past posts on Overcoming Bias have bemoaned neglect of nuclear winter, and high-lighted recent research. Particularly important is a 2007 paper by Alan Robock, Luke Oman, and Georgiy Stenchikov: “Nuclear winter revisited with a modern climate model and current nuclear arsenals: Still catastrophic consequences.” Their model shows severe falls in temperature and insolation that would devastate agriculture and humanity’s food supply, with the potential for billions of deaths from famine in addition to the direct damage.
So I asked Luke Oman for his estimate of the risk that nuclear winter would cause human extinction, in addition to its other terrible effects. He gave the following estimate:
The probability I would estimate for the global human population of zero resulting from the 150 Tg of black carbon scenario in our 2007 paper would be in the range of 1 in 10,000 to 1 in 100,000.
I tried to base this estimate on the closest rapid climate change impact analog that I know of, the Toba supervolcanic eruption approximately 70,000 years ago. There is some suggestion that around the time of Toba there was a population bottleneck in which the global population was severely reduced. Climate anomalies could be similar in magnitude and duration. Biggest population impacts would likely be Northern Hemisphere interior continental regions with relatively smaller impacts possible over Southern Hemisphere island nations like New Zealand.
Luke also graciously gave a short Q & A to clarify his reasoning:
Q1: What food sources would you expect to sustain surviving human populations with severe nuclear winter? The months of existing grain stocks? Slaughtering livestock herds? Intensive fishing? Electric greenhouse agriculture? Simply less-effective agriculture?
A: My thought was that food sources would be mainly fishing as well as less-effective agriculture, assuming little or no access to fertilizer or fuel.
Q2: If nuclear arsenals become much larger in the future, e.g. 100x as large, damage would presumably scale sublinearly (only so many cities to ignite). Could the detonation of millions of nuclear weapons make a material difference to your estimate?
A: Yes it would make a difference but as you state I would definitely think it would scale sublinearly. The largest thing that I would think, more so than the number above a certain point, would be how much the Southern Hemisphere is involved. In the 2007 paper scenario it is assuming largely NH mid-high latitude injection so there is likely large difference in black carbon aerosol amounts in the respective hemispheres. This is one of the largest differences between the 150 Tg of BC scenario and that of Toba, which was a tropical eruption and presumably spread much more evenly over both hemispheres.
Q3: Am I right in thinking that the estimate is based on the reasoning that many Toba-level events must have taken place in the last tens of millions of years, but did not wipe out our prehuman ancestors (even if perhaps eliminating some other lineages of hominids, or bringing human ancestor populations near minimal sustainable size), so the probability per event must be low (plus our access to modern technology)?
A: Yes that was my thinking.
Q4: 1 in 10,000 to 1 in 100,000 is quite a low probability, although one that could be justified if we were sure that similar events had happened many times. However, it is also low enough for model uncertainty to matter. In particular, how much probability mass can we place in nuclear winter being less or more dangerous than a Toba-level eruption? Should we assign a 1-10% probability in it being materially worse than Toba in terms of human extinction risk? In other words, how fat are the tails of the distribution for nuclear winter climate models?
A: Yes there is definitely plenty of model uncertainty when dealing with these kinds of scenarios. This question sort of goes back to my answer to number 2 in that the impacts would likely be different in the respective hemispheres, with the Northern Hemisphere more likely to be Toba-like in climate impacts. My thought for the extinction question was to treat the Southern Hemisphere as the rate limiting step. So, in the scenario we assumed, the NH climate impacts might have a 20-30% chance of being materially worse but the SH maybe around 1-5% chance of being worse.
Also, I was thinking of something in the range of 1,000-5,000 as the Minimum Viable Population (MVP) but if it is on the high end it could lower my estimated probability somewhat, but probably not significantly. Probably one of the biggest uncertainties on my end is my climate change estimate for Toba. Papers after ours suggest a smaller climate impact due to different aerosol size assumptions than we used. So if indeed there was a population bottleneck around Toba and the climate anomalies were significantly smaller than we assumed, this would likely significantly raise extinction probabilities.
Q5: There are widespread popular claims that nuclear winter would create a significant chance of human extinction. Could you name other climate scientists who would estimate higher probability than yourself?
A: I haven’t really read any accounts where there was a probability placed on human extinction. I certainly could be offbase with my estimate, it is not something I have done before. I don’t know offhand anyone that would estimate higher but I am sure there might be people who would. [I asked two colleagues] who did respond back to me, saying in general terms “very close to 0” and “very low probability.”
I guess a big question that I don’t really know or have a good feel for is how big is the discontinuity between something impacting 99% of the population and that of 100%. In my thinking about this problem I thought it might be quite large but maybe it is not. It also probably makes a big difference if the forcing is a pulse that goes away after several years or a constant forcing.
a WordPress rating system