Intermediate Report on Abrupt Sunlight Reduction Scenarios

This project by CEARCH investigates the cost-effectiveness of resilient food pilot studies for mitigating the effects of an extreme climate cooling event. It consisted of three weeks of desktop research and expert interviews.

Headline Findings

  • CEARCH finds the cost-effectiveness of conducting a pilot study of a resilient food source to be 10,000 DALYs per USD 100,000, which is around 14× as cost-effective as giving to a GiveWell top charity[1] (link to full CEA).

  • The result is highly uncertain. Our probabilistic model suggests a 53% chance that the intervention is less cost-effective than giving to a GiveWell top charity, and an 18% chance that it is at least 10× more cost-effective. The estimated cost-effectiveness is likely to fall if the intervention is subjected to further research, due to optimizer’s curse[2].

  • The greatest sources of uncertainty are 1) the likelihood of nuclear winter and 2) the degree to which evidence from a pilot study would influence government decision-making in a catastrophe

  • We considered two other promising interventions to mitigate the effects of an ASRS: policy advocacy at the WTO to amend restrictions that affect food stockpiling, and advocating for governments to form plans and strategies for coping with the effects of Abrupt Sunlight Reduction Scenarios (ASRSs). We believe these could possibly be very cost-effective interventions, although we expect them to be difficult to assess.

  • Our detailed cost-effectiveness analysis can be found here, and the full report can be read here.

Executive Summary

In an Abrupt Sunlight Reduction Scenario (ASRS) the amount of solar energy reaching Earth’s surface would be significantly reduced. Crops would fail, and many millions would be without food. Much of the research on mitigating the effects of these catastrophes is the product of ALLFED. Their work has been transformative for the field, but may be seen by some as overly pessimistic about the likelihood of agricultural shortfalls, and overly optimistic about the effectiveness of their suggested interventions. This project aims to provide a neutral “second opinion” on the cost-effectiveness of mitigating ASRSs.

Nuclear winter is probably the most likely cause of an ASRS, although the size of the threat is highly uncertain. Nuclear weapons have only once been used in war, and opinions differ on the likelihood of a future large-scale conflict. Even if there is a major nuclear war, the climate effects are contested. Much of the scientific literature on nuclear winter is the product of a small group of scientists who may be politically motivated to exaggerate the effects. Critics point to the long chain of reasoning that connects nuclear war to nuclear winter, and argue that slightly less pessimistic assumptions at each stage can lead to radically milder climate effects. We predict that there is just a 20% chance that nuclear winter follows large-scale nuclear war. Even so, this implies that nuclear winter represents over 95% of the total ASRS threat.

There are a number of ways to prepare: making plans; fortifying our networks of communication and trade; building food stockpiles; developing more resilient food sources. After a period of exploration, we decided to focus on the cost-effectiveness of conducting a pilot study for one resilient food source. “Resilient” food sources, such as seaweed, mass-produced greenhouses, or edible sugars derived from plant matter, can produce food when conventional agriculture fails, mitigating the food shortage. We know that these sources can produce edible food, but none have had pilot studies that identify the key bottlenecks in scaling up the process rapidly in a catastrophe. We believe that such a pilot study would increase the chances of the resilient food source being deployed in an ASRS, and that the lessons learnt in the study would enable the food source to be harnessed more productively.

We assume that resilient foods would not make a significant difference in milder scenarios, although this is far from certain[3]. We do not attempt to measure the benefits of resilient food sources in other climate and agricultural catastrophes[4].

Unlike previous analyses, ours accounts for specific reasons that resilient food technologies may not be adopted in a catastrophe, including disruptions to infrastructure, political dynamics or economic collapse. We attempt to model the counterfactual effect of the pilot itself on the deployment of the food source in a catastrophe. Due to lack of reliable data, however, we rely heavily on subjective discounts.

We estimate that a USD 23 million pilot study for one resilient food source would counterfactually reduce famine mortality by 0.66% in the event of an ASRS, preventing approximately 16 million deaths from famine.

Mortality represents 80% of the expected burden of an ASRS in our model, with the remainder coming from morbidity and economic losses. There is some uncertainty about the scale of the economic damage, but we are confident that famine deaths would form the bulk of the burden. We do not consider the long-term benefits of mitigating mass global famine.

Our full CEA assesses the intervention in detail. We draw upon objective reference classes when we can, and we avoid anchoring on controversial estimates by using aggregates where possible.

Our final result suggests that there is a distinct possibility that the cost-effectiveness of developing a resilient food source is competitive with GiveWell top charities. The result is highly uncertain and is especially contingent on 1) the likelihood of nuclear winter and 2) the degree to which evidence from a pilot study would influence government decision-making in a catastrophe.


Link to full CEA.

Overall, we estimate that developing one resilient food source would cost approximately USD 23 million and would reduce the number of famine deaths in a global agricultural shortfall by 0.66%. We project that the intervention would have a persistence of approximately 17 years.

Given that we estimate the toll of a global agricultural shortfall to be approximately 98 billion DALYs, we obtain an estimated cost-effectiveness of 10,000 DALYs per USD 100,000, which is 14× as cost-effective as a GiveWell top charity[1].

The final cost-effectiveness calculation is approximately

With the following figures:

98,000,000,000Burden of an ASRS, DALYs
0.87Proportion of the burden that is affectable once the delay in conducting the study is accounted for
0.00025Annual probability of an ASRS
0.0066Reduction in burden due to resilient food pilot study
17Persistence of intervention (equivalent baseline years)
23,000,000Cost of intervention, USD

The calculation above is heavily simplified. Check the full CEA to see how the figures above were reached.

Overall uncertainty

Although the cost-effectiveness is estimated to be 14× that of a GiveWell top charity, our uncertainty analysis suggests there is only a 47% chance that the cost-effectiveness is at least 1× that of a GW top-charity, and a 18% chance that it is at least 10×. Hence the central cost-effectiveness estimate is heavily influenced by a minority of “right-tail” scenarios of very high cost-effectiveness.

We derived the above estimate by creating an alternative version of our CEA that incorporates uncertainty. Most input values were modeled as Beta or LogNormally-distributed random variables, and the adapted CEA was put through a 3000-sample Monte Carlo simulation using Dagger. In most cases the input distributions were determined subjectively, using the same mean value as the point-estimate used in the CEA. An important exception is the cost: instead of modeling cost as a distribution and dividing “effectiveness” by “cost” to get cost-effectiveness, we model the reciprocal of cost as a distribution, and multiply “effectiveness” by this to get cost-effectiveness. This allows us to avoid obtaining a different central estimate to our CEA due to the result E[X/​Y]≠E[X]/​E[Y].

Link to full report //​ Link to cost-effectiveness analysis //​ Link to summary of expert interview notes

  1. ^
  2. ^

    The headline cost-effectiveness will almost certainly fall if this cause area is subjected to deeper research: (a) this is empirically the case, from past experience; and (b) theoretically, we suffer from optimizer’s curse (where causes appear better than the mean partly because they are genuinely more cost-effective, but also partly because of random error favoring them, and when deeper research fixes the latter, the estimated cost-effectiveness falls).

  3. ^

    In mild agricultural shortfalls such as those that may be triggered by crop blight, VEI-7 volcanic eruption or extreme weather, adaptations like redirecting animal feed, rationing and crop relocation would in theory be sufficient to feed everyone. However, it’s plausible that resilient foods could ease these crises by filling gaps in available nutrition (protein, for example) or providing new sources of animal feed which make more crops available for human consumption.

  4. ^

    Resilient food sources could prove useful in a variety of agricultural shocks. Industrial sources such as cellulosic sugar plants could form a reliable food source in scenarios of 10+°C warming, or in the face of an engineered crop pathogen targeting the grass family [credit: David Denkenberger]