Hide table of contents

On this forum, I have referred several times to a paper showing striking cost-effectiveness of getting prepared for global agricultural catastrophes. It is now published here. I acknowledge several EAs who reviewed the paper. The abstract is below; we also analyze return on investment and find extremely high values. We do not quantitatively compare to the effectiveness of working on other global catastrophic risks, but because this is such a leveraged opportunity, it is likely to compare favorably. The number of expected lives lost per day delay of getting prepared is what convinced me to give significant fraction of my own money to the effort.

I am interested in your feedback on the assumptions, and also how to communicate the cost-effectiveness to EAs and the general public. The charity we are starting would not only do the direct work to get prepared, but it would also hopefully motivate additional funding. This should be even more cost-effective than the direct interventions, but I would probably be conservative and ignore that. For most audiences, I would also be conservative and ignore far future benefits. Another source of conservatism is that our budget will be small compared to the tens of millions of dollars required to do significant preparation, so we can choose the most cost-effective activities. Much of the preparations for ~10% global agricultural shortfalls would be valuable to prepare for ~100% global agricultural shortfalls (large comet/asteroid, super volcanic eruption, and nuclear winter). Ignoring these benefits is another source of conservatism. There also sources of conservatism that affect overall cost-effectiveness, but not cost per life saved, including preserving biodiversity. We also ignore the reduction of the cost of food during the catastrophe for the people who would have survived anyway. Preliminary calculations indicate that this would make the cost to developed countries (assumed to be the donors) of getting prepared net negative, meaning net negative cost to save expected lives. But I have not yet written that paper, so let's return to the conclusions of the published paper.

The general public typically does not do very well with uncertainty, so I was thinking of using the median value of $10 to save an expected life. I think the media would fixate on the lower bound of saving expected lives for $.30 apiece. Might this be ok because of the large conservatism above?



The literature suggests there is about a 1 % risk per year of a 10 % global agricultural shortfall due to catastrophes such as a large volcanic eruption, a medium asteroid or comet impact, regional nuclear war, abrupt climate change, and extreme weather causing multiple breadbasket failures. This shortfall has an expected mortality of about 500 million people. To prevent such mass starvation, alternate foods can be deployed that utilize stored biomass. This study developed a model with literature values for variables and, where no values existed, used large error bounds to recognize uncertainty. Then Monte Carlo analysis was performed on three interventions: planning, research, and development. The results show that even the upper bound of USD 400 per life saved by these interventions is far lower than what is typically paid to save a life in a less-developed country. Furthermore, every day of delay on the implementation of these interventions costs 100–40,000 expected lives (number of lives saved multiplied by the probability that alternate foods would be required). These interventions plus training would save 1–300 million expected lives. In general, these solutions would reduce the possibility of civilization collapse, could assist in providing food outside of catastrophic situations, and would result in billions of dollars per year of return.

Sorted by Click to highlight new comments since:

I haven't previously engaged with your writing on this topic; I appreciate your calling attention to the promise of this as a cause area, and your persistent, rational engagement with the topic.

First of all, I was thrilled to see an acknowledgment of the inconsistency of the VSL with the underfunding of global charities. I myself have considered writing a paper specifically on this topic and the implications with regard to CEA/CBA.

Second, looking at the paper, it seems that the conclusions in the final table are without discounting future lives saved. If you were to apply discounting, how are your CEA conclusions affected? Would be interested in seeing the sensitivity analysis there.

Third, which component of the 1% risk would you find most questionable / possibly affecting the overall CEA conclusion? I generally buy that number, and that this is a promising cause area, but I'd like to investigate it a bit more myself, and your writing thus far implies that your response would be trustworthy on this.

With regard to your main question in this post: "How to communicate the cost-effectiveness to EAs and the general public. The charity we are starting would not only do the direct work to get prepared, but it would also hopefully motivate additional funding."

You are doing the right things to communicate to EAs. You are taking a statistical, skeptical, researched approach. The next steps would just be to make a stable, promising organization in this space with a solid plan for people to engage with and then move beyond just the forum and work on directly engaging with the meta charities and EA donors yourself.

I doubt that communicating with the general public should be a goal of yours. Instead, you most likely intend to communicate with those who are outside the EA community but are interested in food security, ag issues, and global risks. Many of them will already be predisposed to agree with your conclusion and many of its components, though they often will not be as quantitatively adept and engaged. Case-studies can be quite appealing to this audience; referring to regional and global food shortages of the past and the inadequacy of current preparation can be compelling.

In general it's easier to provide feedback on established messaging, such as a website, fundraising prospectus, or mission statement, which I'd be happy to do. Will also be happy to discuss future plans and steps for moving forward if helpful; I do believe I could help with regard to organizing, founding, and messaging.

Thanks for the feedback and your careful read!

Second, looking at the paper, it seems that the conclusions in the final table are without discounting future lives saved. If you were to apply discounting, how are your CEA conclusions affected? Would be interested in seeing the sensitivity analysis there.

We say: "This is because future lives saved are typically not discounted, and the number of lives saved per year would likely increase because of population growth." I think one reason that lives are not typically discounted is that the value tends to grow with GDP per capita, and many say we should only discount at the GDP per capita growth rate (this is valuing utility equally, with a logarithmic utility function). If one did discount lives at a greater rate than GDP per capita growth, it would reduce cost effectiveness somewhat, but we are focusing on the risk in the next couple decades, so it would not be a dramatic change.

Third, which component of the 1% risk would you find most questionable / possibly affecting the overall CEA conclusion? I generally buy that number, and that this is a promising cause area, but I'd like to investigate it a bit more myself, and your writing thus far implies that your response would be trustworthy on this.

The most important uncertainty in the risk is for the multiple breadbasket failure, because that is the largest estimated number. I think the UK government report is reliable, but there may very well be differing opinions in the climate/extreme weather community. However, I did not even try to quantify the number of risks, including regional nuclear war (for example, India-Pakistan), complete global loss of bees as pollinators, a super crop pest or pathogen, a conventional world war or pandemic that disrupts global food trade (and the resultant famine caused in food-importing countries), etc. So even if the multiple breadbasket failure scenario is significantly less likely than the report indicates, I still think order of magnitude 1% probability per year of a 10% global agricultural shortfall is defensible.

The general public may indeed not be a good place to look for funding. However, the more people that know about these solutions ahead of time, the less likely people are to panic and the more likely they are to cooperate in a catastrophe.

Thanks for your offer of additional feedback and help. Our initial website is here.

Interesting paper, I'm glad you're looking in to this!

I took a look, here is some feedback.

Thus, conventional approaches to a 10% global agricultural shortfall would not be adequate to stop an escalation in current hunger-related disease and death (UNICEF 2006).

I had a hard time finding this citation, is there any chance you could quote the relevant section?

To evaluate this claim, we need to know the short-run price elasticity of supply for calories. In other words, how much can the supply of food respond over the interval of time it would take for stockpiled food to run out. My guess is that it would take at least another whole paper to answer this question.

I was doing background research on this, and it looks like just recently a paper was published on the topic of whether Africa can feed itself. Press coverage. Looks like Africa could dramatically increase its crop yields by investing in existing agricultural technology. If these efficiency improvements aren't implemented, Africa will likely see widespread deforestation (due to cropland expansion) as its population grows 2.5x in the next 34 years.

I don't get the impression that population growth is going to slow down after 2050, either. Here are some relevant links:

(I highly recommend these articles. The more I read about this topic, the more confident I get that family planning and African women's empowerment should be a top EA cause area.)

Anyway, this article has a quote from an agriculture expert:

After international food prices soared in 2008, both donors and African governments invested more in agriculture, but investments fell once the prices leveled off, he said.

In your paper, you write that a 10% global agricultural shortfall would "roughly triple the price of grain". Looks to me like this is roughly what happened in 2008.

You mention the possibility of price speculation. I think this sort of speculation is generally a good thing, because it creates incentives to increase food production in advance of a crunch. I know in the US we have anti-price-gouging laws that work against this incentive; I'm not sure if these laws exist in other countries. If so, it might be wise to work on repealing them.

Food price increases are easier to solve than food shortages, because philanthropists can, at least theoretically, step in and cover the price differences for the world's poorest. (Which apparently already happened in 2008?)

Many of these alternate food solutions, if ramped up quickly to cover a 100 % agricultural shortfall individually, could be much more expensive than the new high grain price. However, in a 10 % agricultural shortfall, each food source would only have to provide a relatively small amount of food.

I would expect economies of scale to apply. Probably we'd be best off concentrating all our energy on a single alternative food source and trying to make it as cheap as possible.

If alternative food sources could be produced as cheaply per calorie as grain, I'm surprised entrepreneurs haven't already started companies to commercialize them. Maybe this is something you could do? In general, I think if you have the choice between starting a for-profit and a non-profit, the for-profit is a better option because it lets you make money and donate to other causes.

Speculation: the key issue for an alternative food source is the short-run price elasticity of supply--how responsive is supply to price changes? For a crop that has a long growing cycle, the supply can't respond very quickly. So the question is: can we invent a new food source that's cheap and can be ramped up quicker than any existing source?

But I'm not sure any of this even matters if current crop growing cycles are shorter than the length of time it would take for stored food to run out.

Anyway, it might be worthwhile to get economists and agriculture experts to give you more feedback. Doing so will hopefully increase the chance that your ideas are taken seriously by the "West Coast smart-philanthropy set".

Thanks for the feedback!

We are using the UNICEF quote for the current number of people who die from undernutrition related causes. There was indeed a significant price run-up in 2008, and that caused greater malnutrition. But the actual shortfall was only around 1% of global agricultural production. So our point is that if we had a 10% global agricultural shortfall, the situation would be much worse. When we say triple, this would mean based on current higher prices, so the overall price would be much higher. We do not have a direct quote saying that conventional measures would not be able to handle a 10% shortfall. However, talking with places like the World Food Program, they are struggling with current crises and really cannot imagine a 10% shortfall.

Price speculation is a double-edged sword. I agree that higher prices are good in developed countries because it spurs production and discourages waste. It can also be good for poor farmers, but generally not the poorest who are subsistence (and not selling their food). However, if this prices the global urban poor out of food, that is not good. I agree that if philanthropists step in to subsidize the food price for the poor, then that would be the best scenario.

I generally do not expect that alternate foods would be lower cost than grain is now. The one possible exception is turning agricultural residues into sugar with enzymes, because in order to be competitive as a fuel, the price needs to be similar to grain (because we currently turn a lot of grain into fuel). But I think for the 10% shortfall, it is more about alternate feed. So this could mean feeding agricultural residues to cellulose digesting animals such as cows, sheep, and goats. This could also mean municipal collection of food waste to feed two pigs. I don't expect these things to be economical on a large scale now, but if grain prices triple, we would have a shot. There may very well be some avenues that are cost-effective now, but I'm guessing existing companies would be better positioned to take advantage of them.

Yes, economies of scale would apply and so would learning based on cumulative production. But I don't know if these factors are more important than the factor of high cost associated with rushing to produce a lot of food quickly. For instance, if we are retrofitting existing industrial processes, we would choose first of those processes with the least opportunity cost.

You are correct that tree-based crops would be slower to respond to this price changes. It is plausible that conventional ramp-up of crops could be done fairly quickly, but food price remained fairly high from 2007 to 2014, until we finally caught up with demand. So basically if we have more options to ramp up food supply, this would reduce the price. How much is important future work.

It was interesting to see you mention ocean fertilization. I know this has been proposed as a solution for global warming as well. It seems like scientists are mostly against it on precautionary principle grounds, which is frustrating. If we just had better ocean property rights, fishing companies might be incentivized to fertilize, which could improve food security while helping to reverse global warming as a side effect. It definitely seems to me like a concept that deserves further study. The best argument against it probably involves wild animal suffering.

Ocean fertilization is another alternative food that might be economical right now, though only with the correct property rights as you point out. In areas of the ocean with low nutrients, many trophic levels including zooplankton are required to get to fish. However, where we fertilize, we could go directly from algae to fish (as happens for natural upwelling). So if you are worried about "insect" (actually arthropod) suffering, ocean fertilization could be better than existing ocean ecosystems (I have made this point to Brian, but it did not end up in his essay).

Interesting. :) Do you have further reading on this point?

It seems that increased phytoplankton in lakes and rivers generally leads to more zooplankton. Do you think the dynamics are different in the oceans? I have a hard time believing that herbivorous fish could not only eat all the extra phytoplankton from fertilization but even some of the phytoplankton that was present pre-fertilization (which is what's necessary to reduce zooplankton populations relative to pre-fertilization levels), but I could be wrong!

Thanks for the information on freshwater systems. I believe the quote about saltwater systems was in this book.

Yeah, successful ocean fertilization is a scary scenario in my eyes due to increasing oceanic animal suffering. A bit more discussion here.

I would expect that improvements to global stability make ocean fertilization a positive prospect in the long run. Do you disagree?

I'm now more ambivalent about global stability than I was previously because, in addition to making uncooperative/violent futures less likely, it also makes space colonization more likely. The overall impact on suffering from a negative-utilitarian standpoint is unclear.

This perspective also requires assuming that the far future dominates over short-term effects in the calculation, which not everyone agrees with.

Another complication: An unstable world could cause human extinction and create the opportunity for some other intelligent species to arise. Even the most destructive human war would probably not kill every animal living around deep-sea vents. We went from the first animals to humans in 600 million years of evolution, and life on Earth probably has at least 1 billion years left (casual Googling). So a new intelligent species evolving on Earth after human extinction seems like a strong possibility.

It's an open question whether we would make for better star colonizers than some later species. Relevant post of mine.

I wish we'd get more systematic about identifying and resolving crucial considerations of this sort.

Thanks. :) I discuss that a bit here. I'd be curious to know your probability that non-humans would re-establish civilization if humans went extinct.

I'd be curious to know your probability that non-humans would re-establish civilization if humans went extinct.

Uninformed speculation follows.

On the face of things it seems pretty likely. "...if the dinosaurs hadn't been killed by an asteroid, plausibly they would still rule the Earth, without any advanced civilization." I got the impression that the dinosaurs experienced several mass extinctions, and mammals displaced them when there was a mass extinction associated with climate change? Periodic mass extinctions are evidence against Earth getting "clogged" this way.

I don't feel like I have a good sense of likely causes of human extinction. Destruction of human civilization seems likely; most civilizations that have existed have eventually ended. But when I look at Wikipedia's page on human extinction, scenarios where every last human dies while other life persists on Earth don't seem super numerous. For example, it seems tricky to engineer a virus with a 100% kill rate that is also infectious enough to infect all 7 billion of us. (Do we have recorded instances of entire species being wiped out due to illness this way?) And if nanobots or some physics experiment eats the planet, that will destroy all the other life too. The most likely scenario seems like destruction of current human civilization alongside destruction of viable ecological niches for technologically unsophisticated human bands--runaway global warming or nuclear winter?

If that's the scenario that comes about, I would guess that lots of animals will survive, analogous to extinction events that killed off dinosaurs. I don't think that a big fraction of the great filter is between development of animals and civilization, although it seems plausible that there is some filter here. A naive way to estimate: divide the number of times civilization has arisen (once) by the number of times Earth has been "wiped" by mass extinctions. Then figure out how frequently mass extinctions occur and how many more "wipes" we can expect before Earth is uninhabitable.

On balance it's plausible our hypothetical replacements would be less compassionate, because compassion is something humans value a lot, while a random other species probably values something else more. The reason I'm asking this question in the first place is because humans are outliers in their degree of compassion.

Why do you believe that humans are outliers in their degree of compassion relative to other social species?

Almost by definition, a species that creates a civilization is capable of large-scale cooperation. But this large-scale cooperation could look much different than human cooperation looks like. (I'm guessing it would be relatively easy for a eusocial species to control its reproduction, so if it achieved sufficient intelligence to understand the basics of breeding, it might be able to "bootstrap" itself to higher levels of intelligence from there.)

(I can imagine exotic scenarios where large-scale cooperation is less necessary for starfaring: consider a species that lived much longer than humans, meaning individuals had longer lifetimes over which to accumulate knowledge, which makes knowledge-sharing through culture less necessary. But I believe that species tend to be longer-lived in highly stable environments, and a highly stable environment is less likely to stumble across a configuration that creates an ecological niche for a highly intelligent tool-using species.)

It occurs to me that we might want to focus on how cohesively a species cooperates over how compassionate it seems to be. If you look at human actions like factory farming, these seem to be less a product of some human prediliction for cruelty, and more a result of incentive structures. In a post-scarcity society, we'd expect this to be less of a consideration. But a post-scarcity society requires more than just technology. Incentive structures also seem less contingent on biological factors and more contingent on societal factors.

Multipandemic could cause human extinction. Even a single virus has had 100% kill rate.

Interesting paper! I'm intuitively skeptical, though--with 7 billion people, it just seems really hard to kill off every last person.

Where was this paper posted?

Sorry-I guess the review period for the paper on academia.edu expired. But contact Alexey: https://fromhumantogod.wordpress.com/contacts/ if you want to see the paper.

Interesting about dinosaurs. :) I added this to my piece as footnote 4 (see there for the hyperlinks):

John Maxwell disputes this claim. My reasoning is that dinosaurs lasted for at least 135 million years but only went extinct 66 mya. It's easy to imagine that dinosaurs might have lasted, say, twice as long as they did, in which case they would still rule the Earth today.

I agree that, among ways all humans might go extinct that leave around other animals, runaway climate change and nuclear war are contenders. However, I'm skeptical that there would be many vertebrates left in these scenarios. I would think humans would try to eat every mammal, bird, and fish they could find. I guess humans couldn't kill every last mouse or minnow, but in these scenarios, plant growth would also be compromised (or else humans would be eating plants), so it's not clear how well these small animals could survive.

A naive way to estimate: divide the number of times civilization has arisen (once) by the number of times Earth has been "wiped" by mass extinctions.

Interesting. :) But this is (perhaps very strongly) biased by the selection effect that we find ourselves here and not on the (many?) other planets where intelligent life never got to the human level.

Why do you believe that humans are outliers in their degree of compassion relative to other social species?

This was a statement about selection bias (namely, that humans care more about what they care about than other civilizations do) rather than an empirical generalization from biology. I agree humans seem to be somewhere in the middle in terms of peacefulness relative to primates. Unlike cetaceans, wolves, etc., we're not obligate carnivores. We're also more compassionate to outgroup members than ants are, although it's hard to say what ant morality would look like if ants were as intelligent as mammals.

I still don't see why we should expect these future extinction events to wipe out all vertebrates when vertebrates made it through dinosaur extinction events. Most plants aren't human-edible, and I'm skeptical humans would be systematic enough in foraging through remote wilderness to kill off more than half the wild vertebrates on the planet.

Interesting. :) But this is (perhaps very strongly) biased by the selection effect that we find ourselves here and not on the (many?) other planets where intelligent life never got to the human level.


humans seem to be somewhere in the middle in terms of peacefulness relative to primates

How certain are you about this? Some quick Googling:

  • "The northern muriqui has been argued to be important to understanding human evolution, since it is one of the few primates that has tolerant, nonhierarchial relationships among and between males and females, a feature shared with hunter-gatherer humans, but which contrasts with the ranked relationships of most other primates." (source)

  • "Male primates, in general, take very little interest in helping to rear offspring... Pair bonding of any sort is rare among primates, though gibbons seem to be lifelong monogamists, and some new world monkey groups, such as marmosets, have only one reproductively active pair in any group." (source)

  • "This was also studied in rhesus macaques and pigtail macaques. They found that infants, when separated from their mothers, went though all these stages of separations- protest, despair etc. The saw the same thing with rhesus and pigtails, but in bonnet macaques, the infants don't go through all this psychological trauma. It's pretty clear why if you look at their social organization- there are a lot of allomothers in bonnet macaques and babies are often left by their moms in the wild and someone else will take care of it and bring it back to her later. So it's important to pick more than one species and to compare across species when you're doing this comparative approach for behavioral models... We also are different because we (both sexes) cooperate with non-kin pretty often." (source--first bit is interesting because I don't think humans really alloparent, which seems like an altruistic behavior?)

  • Random related thought: Somewhere I read that humans "self-domesticated" over the course of our species through e.g. capital punishment for murderers. Does that mean that we are "just cooperative enough" to be civilized? (In other words, did this "self-domestication" process occur until the point at which large scale civilization became possible, and that's where we are right now?)

Some of this stuff might be related to the evolution of intelligence though, e.g. human babies are born prematurely relative to other species because our large heads would not fit through the birth canal otherwise. So perhaps a primate species would need to engage in pair bonding in order to make this sort of 'premature' birth (and thus the evolution of high intelligence) possible. This factor seems relatively contingent on primate anatomy. So maybe a non-primate-descended intelligent species would be less likely to experience pair bonding (I think it's rare in the animal kingdom) and thus be less benevolent. (BTW, I think species that pair bond are a strict (and small) subset of species that are considered K-selected, but I could be wrong. It seems pretty likely that intelligent aliens would be K-selected in some form.)

Lives saved is a very very weird and mostly useless metric. At the very least try and give an estimate in QALYs (quality adjusted life years) since very few people actually value saving life per say (e.g. stopping someone who is about to die of cancer from dying a few minutes earlier).

Given that many non-deaths from food scarcity are probably pretty damn unpleasant this would probably be a more compelling figure.

I agree that QALYs are more robust, and I guess it was an earlier version of the paper where we noted that using QALYs would likely produce similar comparison of cost-effectiveness to global poverty interventions. But we wanted to keep this analysis simple, and most people (though perhaps not most EAs) think in terms of saving lives. Also, two definitions of a global catastrophic risk are based on number of lives lost (I believe 10 million according to the book Global Catastrophic Risks and 10% of human population according to Open Philanthropy).

That is good to know and I understand the motivation to keep the analysis simple.

As far as the definition go that is a reasonable definition of the term (our notion of catastrophe doesn't include an accumulation of many small utility losses) so is a good criteria for classifying the charity objective. I only meant to comment on QALYs as a means to measure effectiveness.

WTF is with the votedown. I nicely and briefly suggested that another metric might be more compelling (though the author's point about mass appeal is a convincing rebuttal). Did the comment come off as simply bitching rather than a suggestion/observation?

I did not do the vote down, but I did think that calling lives saved a mostly useless metric was a little harsh. :-)

Curated and popular this week
Relevant opportunities