Robert Wiblin: Hi listeners, this is the 80,000 Hours Podcast, where each week we have an unusually in-depth conversation about one of the world’s most pressing problems and how you can use your career to solve it. I’m Rob Wiblin, Director of Research at 80,000 Hours.
Today I interview a fascinating and unconventional inventor, who is likely to be both entertaining and informative for most subscribers.
Before that though, this is the show’s fiftieth episode and it’s going to come out around New Year, so I thought it’s a good moment to take a minute to reflect on the year that’s past and where we’re going.
To be honest with you it’s just been an incredible privilege to be able to run this show. Thanks to our donors, Keiran and I are spared the commercial pressures that would otherwise drive us to produce content that’s superficial, or dumbed-down, or misleading, just to get clicks. Instead we just try to produce episodes that we find sincerely interesting and informative ourselves.
As a result, and thanks to people like you recommending the show to their friends, we’ve progressively been able to build up an incredibly highbrow audience. Many of you I meet or get messages from, and the feedback shows both how much you enjoy listening, and the very close attention many of you are paying to each of these interviews.
We’ve produced 95 hours of the show so far, and featured 36 guests in 2018.
I can see from our analytics which episodes you’re most excited to download, and which ones you stick with listening. Whenever Keiran and I worry an interview might have gone too deep into a topic, or be too complicated to hold people’s attention, we’ve found the exact reverse to be the case.
If there’s a level of intellectual challenge that’s too much for you all, we haven’t hit it yet, so I hope next year we can continue to have ever more sophisticated conversations about these topics we all care about, and now with the knowledge that we’ve covered a lot of the basics already.
The second reason we appreciate all of you listeners is your sincere commitment to improving the world. Thanks to our annual impact survey, I’ve been able to read about how many of you are considering totally changing your career plans, or have already done so, on the basis of what you’ve learned listening to this show. It’s an honor to be in the position to help some of the world’s smartest and most caring people have more social impact with their lives.
It’s also a relief to know that this show is not only enjoyable to people, but is making a real contribution to solving the sometimes horrifying problems we talk about here.
I hope that you’ll stick with us into 2019, and that Keiran and I will continue to find guests that can entertain, inform and inspire you.
Alright, with that out of the way, I bring you the eclectic engineer, Dr Dave Denkenberger.
Robert Wiblin: Today I’m speaking with Dr. David Denkenberger. He did his undergraduate in engineering science from Penn State before doing a master’s at Princeton in mechanical and aerospace engineering and a PhD at the University of Colorado Boulder in their building systems program. He’s now an assistant professor at the University of Alaska Fairbanks in mechanical engineering. He co-founded and directs the Alliance to Feed the Earth in Disasters, otherwise known as ALLFED, and donates half of his income to it. He’s authored or coauthored 94 publications with over 1,300 citations, including the book, Feeding Everyone No Matter What, Managing Food Security After a Global Catastrophe. His food security work has been widely covered by the media, including in Science, Discovery Channel Online and Gizmodo, but most importantly of all, he’s a regular listener to the 80,000 Hours podcast. Thanks for coming on the show, Dave.
Dave Denkenberger: Hi, Rob. Thanks for having me.
Robert Wiblin: We plan to talk about the case for and against working on your approach to feeding the world in the case of a major catastrophe. And I know that at the top of the show, you wanted to clarify and disclaim that all of the things you say here are just your opinion, and not the official views of ALLFED, or the Global Catastrophic Risk Institute, or any other group that you’ve been working with. First though, just describe for listeners what ALLFED actually does and why you think it’s such important work.
Dave Denkenberger: The background is that there are a number of catastrophes that could disrupt agriculture globally, and the most extreme ones could basically collapse agriculture. These are the ones that could block the sun, so they include a large asteroid that caused the extinction of the dinosaurs, a super volcanic eruption like many people think almost caused the extinction of humans about 70,000 years ago, and a nuclear winter and this would be caused by what’s called full-scale nuclear war involving thousands of nuclear weapons between the US and Russia. This causes burning of cities and smoke goes into the atmosphere and it can remain there for a decade.
Dave Denkenberger: In these scenarios, since almost all of our food comes from plants that need the sun to grow, it’s generally assumed that most people would die. We only have a few months of food storage. People who have looked at this problem before have suggested, well, let’s just store up more food, and that would be technically possible, but as a way of visualizing it, you can think of a 160-liter or a 40-gallon oil drum. If that’s full of dry food, that can feed a person for a year, but to have enough food to feed seven billion people for five years, you would need to pile those drums from the earth to the moon and back 40 times.
Dave Denkenberger: As you can imagine, that would be very expensive, many trillions of dollars, and you can’t do it really fast. It would take a while to store up that food, but you would want to do it as fast as practical so that you’re ready to weather disasters that could happen at any time. If you store it up fast, then you inflate the food price and then millions of people more would die of malnutrition than already occurs.
Dave Denkenberger: I wasn’t satisfied with that not very good solution. I was thinking, well, is there another way of producing food when the sun is blocked? In 2011, I was reading this paper called Fungi and Sustainability, and the premise was that after the dinosaur killing asteroid, there would not have been sunlight and there were lots of dead trees and so mushrooms could grow really well. But its conclusion was that maybe when humans go extinct, the world will be ruled by mushrooms again. I thought, why don’t we just eat the mushrooms and not go extinct? I branched out from mushrooms, thinking of all the ways that we could convert either dead vegetation into food or even fossil fuels into food.
Robert Wiblin: Is anyone else working on this? Has anyone else had this idea? It seems obvious in retrospect.
Dave Denkenberger: Well, interestingly, Carl Shulman at Future of Humanity Institute did a blog post as soon after I thought of these ideas, before I published the first paper and the book. Yeah, he was thinking that route. Other than that, it seems like it really hasn’t been thought about.
Robert Wiblin: Yeah. I just love this idea. Whether it’s practical or not, I’m just so fond of it because you’ve taken a problem that most people have just thought of as inevitable and basically accepted that if there’s a nuclear war, then maybe billions of people would die of starvation. I guess bringing an engineering mindset that you’ve been trained in, you’ve just seen this as a practical challenge to overcome and pretty quickly come up with a bunch of seemingly plausible ways that we could actually just feed everyone even if the sun were completely blocked out, which is amazing.
Dave Denkenberger: Yeah, that’s right. I do think that that engineering mindset is really important in that certainly, it’s important to know what the problems are and ideally prevent them, but realistically, they still could happen. I think that there should be more work done on this resilience end of the spectrum.
Most likely causes of a global food shortage/ likelihood
Robert Wiblin: You’ve written some blog posts and, in fact, I think published a paper now trying to estimate the cost effectiveness of working on this problem. I think you think that if your cost effectiveness is correct, you would be able to save lives by working on feeding people in these catastrophes for between $0.20 and $400 per life, which would be pretty extraordinarily effective if it were true. I suspect that that’s probably a bit too overoptimistic, although the project is probably worth doing anyway, even if it’s not quite as effective as that. I guess we’ll discuss the cost effectiveness analysis later and which parts of it might be too optimistic or not optimistic enough perhaps.
Robert Wiblin: First, maybe let’s just dive into thinking about the nature of the problem that you’re solving and the concrete solutions that you’ve come up with for it. Not all listeners I imagine will be convinced that a mass starvation of humanity is a terribly likely risk. Let’s go through, what are the major possible causes of a global food shortage and how likely might they be?
Dave Denkenberger: I mentioned asteroid and super volcano, but as the existential risk community has pointed out, generally the natural catastrophes can’t be that bad because they can’t be happening every century or we would have been doomed long ago, though of course there is the possibility that we’ve been extremely lucky, which people have tried to adjust for. I think still, even adjusting for that, the natural catastrophes are lower probability.
Probability of asteroids/ supervolcanoes as GCRs
Robert Wiblin: Do we have any estimate of the probability of an asteroid or super volcano that would cause a risk to human food supply?
Dave Denkenberger: Yeah. If you’re talking about blocking the sun then it really is around one in I think 10 million years or 100 million years for an asteroid, and the volcano is more likely, more like one in 100,000 or a million years.
Robert Wiblin: How robust are those estimates? Because I know we’ve had a couple of volcanoes erupt and changed the biosphere during the million years or so that humanity has been around.
Dave Denkenberger: The latest estimate I saw on the super volcanoes was actually coming up with a larger number that they happen more frequently. Apparently it’s difficult to identify further back than a few million years. There’s some uncertainty there. There’s also uncertainty whether the super volcanic eruption would actually block the sun well enough and long enough to cause mass starvation. It’s the same with nuclear war as well. I’ve written a paper that tries to go through all the uncertainties of, well, how many nuclear weapons would be detonated? Which cities would be hit? What fraction of the combustible material actually burns quickly? How much of that turns into smoke? How much of that smoke makes it into the upper atmosphere? How much do the particles block the sun? How much does it impact agriculture? As you can imagine, there’s uncertainty in all of those steps so it adds up.
Robert Wiblin: Yeah. It seems like there’s this question of how bad nuclear winter would be or how likely it is to happen is pretty controversial. Is seems to be people have strong views on both sides. Have you had a chance to really dive into that and form a strong view?
Dave Denkenberger: Well, I did dive into it in the course of writing that paper. Yeah, I think you’re right that there seems to be two camps. One is if there’s full-scale nuclear war, there’s going to be nuclear winter. The other camp is even if there is full-scale nuclear war, there’s not going to be nuclear winter. What I came up with was defining it as global agricultural collapse, which means basically the plants we’re growing now, where they are, are not going to be able to grow. Now we’ll get into whether we might be able to move them or not, but if we just use that definition, then I was getting around 20% chance if full scale nuclear war then you would have global agricultural collapse.
Dave Denkenberger: One reason of mine it was relatively low compared to many analysts was because I considered two scenarios that wouldn’t be quite as bad that might still be described as full-scale nuclear war. One of those is called the industrial strike where you’re trying to disable industry but not trying to kill people. The other one is the counterforce strike where you’re trying to disable the other person’s nuclear weapons. Both of those, you’re not hitting city centers and so the fires would not be as bad.
Robert Wiblin: Do you know what the crux of the debate is about? How is it that people can disagree so much about whether a nuclear winter is likely?
Dave Denkenberger: Well, I think it’s because just there are so many levels of uncertainty. People point out in World War II, one of the cities had a firestorm which is defined as a stationary fire. The entire city burns at once. One of them had a mass fire but didn’t all burn it once. This actually affects how much of the smoke actually goes into the upper atmosphere. That’s one example of uncertainty, but yeah, it’s like the Fermi paradox. When you calculate how many civilizations are in our galaxy, some people have come up with zero and some people have come up with millions. There’s this huge uncertainty. At least in this case, I think the uncertainty can be more like two or three orders of magnitude, and so you can actually say something useful when comparing things.
10% global food production shortfall
Robert Wiblin: What about the more moderate cases where there’s a more normal war or an a regional nuclear war. How much would that interfere with the food supply?
Dave Denkenberger: That’s the class of catastrophes that I’ve been labeling the order of magnitude 10% global food production shortfall. Roughly 3% loss to 30% percent loss. In those scenarios, there are a number of catastrophes that could cause that. As you mentioned, regional nuclear war between India and Pakistan, that might only involve 100 total nuclear weapons and actually much smaller nuclear weapons than the US and Russia have, but still they would be targeted at highly populated cities. Some work on this has estimated something like a 10% or 20% global food production shortfall because of it. You can think of smaller versions of the natural catastrophes like asteroid and super volcano, but again, they’re not too likely.
Dave Denkenberger: Another one that’s received some interest recently is called the coincident extreme weather or multiple breadbasket failure. Here, the scenario is you have droughts or floods on multiple continents at once. There was a UK government study on this that estimated right now, it might be around 1% chance per year, but with the slow climate change, that extreme weather probability, so difference between climate and weather actually gets more likely. They were getting more like 80% chance of this century that something like that would happen.
Robert Wiblin: Wow. Okay.
Dave Denkenberger: There are other scenarios as well, like the superweed, which is not the savior of medicinal marijuana users but is instead a weed that out competes crops. If this were a natural thing, we could probably contain it or at least slow it down so that it wouldn’t happen too fast. If we have more time in any of these, it makes it less extreme, but it could be an actual coordinated terrorist attack. There have been some examples of people trying to use … I think that was more a crop disease, but that’s another category, something that would kill crops directly could be a coordinated attack as well.
Robert Wiblin: A superweed sounds a bit outlandish because how would a weed spread all around the world really quickly? It seems like wouldn’t a super pathogen like a virus or bacteria be more able to spread to lots of crops quickly enough that it’s hard to respond?
Dave Denkenberger: Yeah, right. It’s hard to have something that can live in many different climates, yet I’m particularly worried about on the crop disease. If something could target the grass family, then you affect not just grass that feeds a lot of our animals, but also wheat, corn, maze, rice and sugar cane. If you add up all of the human calories, it’s something like two-thirds of our calories comes from the grass family. That really could be catastrophic.
Risk of terrorism involving superweeds
Robert Wiblin: You mentioned in the book that a coordinated terrorist attack using superweeds to destroy a lot of crops across a very large region could be a global threat that not that many people are thinking about. Do you have any evidence that a group might actually plan such an attack or has been considering planning such an attack and do you know if anyone else is working on trying to figure out if this is a real risk and what might be done?
Dave Denkenberger: I have not heard of evidence for this particular type of risk, though a related one is crop diseases. These have been used in biological warfare programs, so there is concern that this could be a potentially larger attack. As for the probabilities, I haven’t seen anything quantitative. I just try to quantify what I can and then, say, “Well, we’re not including all these other things, so really the risk is higher.”
Robert Wiblin: Are there any just nonnuclear wars that you think would create the scenario that are plausible? Who could end up fighting without nuclear weapons? If you had a major war, wouldn’t that massively interfere with trade, and so some countries that are currently importing most of their food could well end up starving pretty fast?
Dave Denkenberger: That’s right. The other scenario that could really interfere with trade is a pandemic. If it were a severe one, then it may be rational to close borders to reduce transmission, but you’re right, importing food countries would be in big trouble or it could be a smaller pandemic and just people overreact. It might not be rational, but they still might close the borders.
Dave Denkenberger: There are a couple of other scenarios that could do … similar to this idea of a 10% global shortfall. One of them is called abrupt climate change. This is like Europe going back into the Ice Age, which some people have talked about because of breakdown of the thermohaline circulation, circulation of the ocean that’s driven by salt. That could be a 10% shortfall.
Dave Denkenberger: Another one that people have talked about and some existential risk researchers are looking at is extreme climate change. It’s global but it still happens slowly, like over a century. That’s a little different than the problems I’m mainly focusing on, but I think some of the ideas could be relevant to that.
Robert Wiblin: Yeah. People worry about us just not having enough food over the next century because population is growing quickly and maybe food technology won’t keep up. Do you think just a food shortfall, even setting aside any disasters, is also a serious risk?
Dave Denkenberger: Well, I think it is important to look at, but generally, if the shortfall happened slowly, we have time to react. For the smaller shortfalls, you could always say, well, technically, we feed more than 10% of the food we produce to animals now. Why don’t we just feed it to people? Well, yes, that’s true, but in reality, people in developed countries like to eat meat, many of them. I think the food price would go so high that you could easily have hundreds of millions of people starve. That’s a particular concern if it happens suddenly. If we have more time, I think there are other things we can do.
Robert Wiblin: Let’s say that you had this 10% shortfall of food scenario. How bad do you think that would be? Some people would die but the rest of us would carry on?
Dave Denkenberger: I wrote a paper about this as well. Again, very large uncertainty on what could happen. If we had a massive outpouring of philanthropic work, we could really limit how many people die. On the other end, we could not just have hundreds of millions of people dying, but then tensions could be very high, and you could imagine nuclear war breaking out, even full-scale nuclear war. I think it has a pretty fat tail that the impacts could be really bad.
Robert Wiblin: Why would people start a nuclear war, just because there wasn’t enough food?
Dave Denkenberger: Well, if people are actually starving, if only a certain number of people are going to survive, you want the people in your country to survive. It is actually in your interest to have conflict. This is the lifeboat ethic, and this is exactly what I want to get away from with having alternate food supplies so that we could say, “Well, actually, if we cooperate, we can feed everyone.” We don’t have to have this lifeboat ethic.
Robert Wiblin: The idea is you would be happy to fight a war that resulted in the deaths of other people because then there’ll be more food left for you.
Dave Denkenberger: Yeah, that’s the logic.
Robert Wiblin: Even though the war presumably would interfere with agriculture to an even greater degree.
Dave Denkenberger: Well, if it’s nuclear, yes, potentially. I’m not saying this is what I would recommend, but there’s always the possibility of irrational decisions.
Robert Wiblin: Just to wrap up, we’ve got asteroids and super volcanoes that would create a serious problem. Most people would probably agree with that, but they’re quite unlikely each year. We’ve gotten nuclear winter, which seems a bunch more likely, although a bit … We don’t know exactly how likely it is because we don’t have a long historical record, and there’s a bit of uncertainty about how severe the weather effects would be, but we certainly can’t rule that out. Then, we’ve got a whole lot of other more moderate scenarios that create 10% or something shortfall in food output, which then could cascade into other problems, including superweeds or super pathogens or sudden weather changes or just bad luck with the weather or more moderate size wars. Altogether you think this adds up to a significant risk to humanity as a whole, something that we should be worried about.
Dave Denkenberger: That’s right.
Robert Wiblin: Who’s working on this problem? I guess there’s some people who would try to store up food and presumably some countries have food stored away. How much is that and what else are people trying to do?
Dave Denkenberger: There are certain countries that have done a lot of food stockpiling. One of them is Switzerland. I believe they have around a one-year food supply. I think China does as well, though it’s not public knowledge so you don’t know exactly. On average, it’s more like a few months across the world. You’ll often hear numbers of we have two months of grain supply. That’s actually about right, but for the wrong reasons. It’s at current grain consumption, but of course we feed half of our grain to animals. If we did have one of these catastrophes, hopefully we would be not feeding as much edible food to animals so that helps, but there’s also the fact that most of our stored food is grain because it’s really easy to store, dried, and half of our calories comes from food that is not grain. We have less food storage that is not grain. It works out to just a few months of food storage, and it depends on what time of the year, but that’s the minimum.
Robert Wiblin: Currently we get, you were saying, about two-thirds of our calories from grains of various kinds.
Dave Denkenberger: Yeah. That is when you count direct consumption and food going into animals.
Robert Wiblin: The other third is fruits and vegetables and fish and things like that.
Dave Denkenberger: Right. There’s debate about whether soybean is a grain or not, but it’s a staple. Yeah, basically fruits, vegetables and other things, nuts.
Robert Wiblin: Nuts, okay. Oils maybe.
Dave Denkenberger: Yeah.
Robert Wiblin: Is there anyone else who’s trying to work on feeding the world in these disaster scenarios using something other than just storage?
Dave Denkenberger: I’m not aware of anyone else.
Robert Wiblin: It’s just you.
Dave Denkenberger: Yeah. Well, ALLFED. Luckily I have a team.
Robert Wiblin: Let’s move on to talking about some of the solutions to this problem. If we had a major nuclear winter and people hadn’t really planned ahead, any way of dealing with this, and they start running out of food stores, what things might people start doing? What have they done in the past when they’ve been starving?
Dave Denkenberger: Well, it’s been really terrible. I’ve heard stories of people in the I think Irish potato famine boiling boots because it was made of leather and they could get some calories out of them. I’ve heard of people eating grass straight, but it has so much fiber that you actually don’t get any net calories. It’s like eating 100% celery. The options are really limited.
Dave Denkenberger: Related to eating grass, there has been a technique to extract edible calories out of leaves where you grind up the leaves and squeeze out the liquid so that you get rid of the fiber basically that we can’t digest. You boil that liquid and a protein concentrate comes to the surface. What is left over in the liquid is actually a lot of the toxins that aren’t good for us. This has already been done at the small scale and less developed countries and it’s even been done on the industrial scale in France. That’s a next step of making some of the existing biomass directly editable for people.
Robert Wiblin: That doesn’t get you very far.
Dave Denkenberger: Well, you might only get 10% of the calories total in the leaves as human food.
Robert Wiblin: It sounds like … Wouldn’t there be a lot of actual manual labor that would go into digesting it and extracting it? Is it even clear that you’d come out ahead?
Dave Denkenberger: To back up about these catastrophes, many people will assume that if there’s nuclear war then our entire industrial infrastructure would be collapsed. That may very well be true in the countries that are hit directly by the nuclear weapons, but say it’s a US-Russia exchange. That might be 20% of the global industrial infrastructure. From a global perspective, I would say the majority of infrastructure would still be functioning. I think there is potential to retrofit our industrial infrastructure to producing food. I actually use the analogy from World War II where before the war, US was hardly producing any airplanes, but once it entered the war, it retrofitted its automobile manufacturing plants to produce airplanes and tanks and was able to do that in a very short amount of time.
Dave Denkenberger: I think it’s feasible to look at our chemical processing plants and retrofit them to produce food. One of the ways of doing that would be looking at how we produce biofuels right now. We’re all familiar with first-generation biofuels where we turn corn or soybeans into transport fuel, but the second-generation biofuels or the cellulosic biofuels take the corn stalk and break it into sugar with enzymes and then feed that sugar to a fungus to make ethanol. If our problem is not having enough food, it may be possible to eat that sugar directly.
Robert Wiblin: Interesting. It sounds like you’re more focused on the countries that haven’t been directly affected by the disaster because that’s a somewhat more solvable problem. If there’s a nuclear war in the northern hemisphere or hopefully an asteroid hits the northern hemisphere, then you can think about, what could New Zealand and Australia and Chile and Brazil do that could plausibly save almost everyone in those countries?
Dave Denkenberger: Yeah, it’s certainly easier.
Robert Wiblin: Basically, the unifying idea in this book, Feeding Everyone No Matter What, is that even if the sun goes away, there’s actually just a ton of chemical energy stored on the earth’s surface in the form of wood and soil and other materials. The thing is humans can’t eat those, so we have to find some way of converting all of that chemical energy into something that humans can then digest easily. I guess you’re going through various different options for doing that.
Dave Denkenberger: Right. There are quite a few options for the leaves that we’ve talked about, the cellulose digesting animals like cows, sheep, goats, rabbits can digest. We could also, like we said, turn them into sugar in an industrial process. Wood is actually more difficult, but mushrooms can grow directly on wood. You can get mushrooms from wood, and they can soften the wood, and that leftover from the wood can actually be fed to cows, sheep and goats and bison. That’s actually already been done. You’re right. It’s just a matter of turning this big energy resource into something that’s actually human edible.
Robert Wiblin: Do we know what the denominator there is? How much chemical energy is there in in wood and other things like that?
Dave Denkenberger: It’s very large. In the book, I look at what if we were trying to get all of our food from just one of these food sources, which in reality we’d be getting food from multiple sources but just as a first analysis. The only one that we actually would run out of feedstock or energy source was outdoor mushrooms growing on logs that they’re so inefficient that we would actually run out. Many of these, you really could feed everyone for five years.
Robert Wiblin: Wow. Where is all of this biomass? Is it just in forests that I don’t really see?
Dave Denkenberger: Yeah. Contrary to popular belief, there still are forests left and, yeah, it’s I think hundreds of gigatons of biomass.
Robert Wiblin: Let’s maybe go through the main options that you can within the book. We’ve got mushrooms, which can grow on just plant matter. I guess we’ve already talked about cows, but there’s other potential things that could eat these materials like insects and rats. There’s trying to just digest them with enzymes I guess in big vats and cook them effectively. There’s this other method of growing bacteria on fossil fuels or finding a way to grow something that humans can eat using fossil fuels. Let’s maybe just do mushrooms first. Do you want to describe some more of the details of that?
Dave Denkenberger: You can grow mushrooms just on a log outside, but that’s limited because it’s only in the tropics that would not be freezing. I’m basing my work on climate simulation of nuclear winter where the global temperature reduction is around seven or eight degrees Celsius. So large areas, at least in the tropics, it would actually not freeze. So you could potentially grow mushrooms outdoors. It turns out it would be very difficult to ship hundreds of billions of tons of wood from the mid-latitudes outside the tropics into the tropics, so I don’t actually consider that option.
Dave Denkenberger: But we could do some growing of mushrooms indoors. We would probably do it on the leaves because that can be converted to mushrooms very quickly. And so you can get more mushrooms per growing area, and people could potentially have racks of mushrooms in their basement.
Robert Wiblin: We would go out and harvest wood and plants and things like that, and then grow mushrooms on them. Can the mushrooms grow quickly enough, and how much kind of space would you need to these many … I mean mushrooms aren’t that calorie-dense. So it seems like we would have to convert, basically, what … I guess, eyeballing it, it seems like we have to convert an enormous amount of all of our building space and all of our work into just growing mushrooms.
Dave Denkenberger: I think if we were getting all of our food from mushrooms, I might’ve estimated in the book, it would take a third of our building square footage, which would be a huge impact. But again, we’re going to be getting food from a number of different sources, so it wouldn’t be all from mushrooms in our basements.
Robert Wiblin: So this is just one option?
Dave Denkenberger: Yep.
Robert Wiblin: Hold on, so the problem with mushrooms is that you can’t really grow them outside because it gets too cold.
Dave Denkenberger: Outside of the tropics, right.
Robert Wiblin: So for Australia, this isn’t really going to work. Maybe it would work in Indonesia or something.
Dave Denkenberger: Yeah.
Robert Wiblin: Are there any other weaknesses of the mushroom approach?
Dave Denkenberger: Well, as you said, they’re not very calorie-dense. They’re actually pretty high percent protein by calorie, but there is a lot of water. But it would be feasible to get enough calories if you ate enough mushrooms. But I did write a paper on the nutrition of alternate foods, and I found to meet the United States recommended daily allowance, you would need a mixture of, basically, all 10 of these foods.
Dave Denkenberger: But to do just what people could survive on, like not get scurvy, I think you could have a more limited mixture of foods. I just haven’t done that analysis.
Robert Wiblin: Do you think it would be worth doing an experiment where you get someone to only eat mushrooms for a couple of months and see what happens?
Dave Denkenberger: I think that would be very risky. I think we say in the book that if you ate a variety of alternate foods, but something that might actually be a diet that we could produce if the sun were blocked, and get permission from your doctor and take a multivitamin, then yeah. I think that would be good to know.
Dave Denkenberger: About the vitamins, one way of getting sufficient vitamins is having a variety of foods. But of course, the people who are poorer would typically not get the variety of food, which is what happens today. And so there are backup plans for vitamins like growing bacteria that would have a particular abundance of vitamin, or maybe even [chemical] synthesis. I haven’t analyzed that, because I found that it would be feasible, if you had a variety of food to get enough vitamins.
Robert Wiblin: How do you think people would react to having to eat such a large amount of unpalatable food? Do you think people would just accept it because the alternative is death?
Dave Denkenberger: I think most people would, maybe not the foodies.
Robert Wiblin: Do you know what volume of mushrooms you would have to eat roughly to survive? I mean because it seems like even many of these other foods that you’re suggesting wouldn’t be as calorie-dense as the foods that we normally eat.
Dave Denkenberger: Well, in the case of mushrooms, many of them are around 90% water. The easy calculation, at least for people in the US, is you need about a pound of dry food per day, so half a kilogram. If it were all mushrooms, it would be about five kilograms. But other ones could be significantly less water like if we were eating cow meat, for instance.
Robert Wiblin: Maybe let’s just back up and … Why can’t we use cows and sheep and other animals that eat these plants as the main food source?
Dave Denkenberger: The basic problem is that these large mammals only have about one offspring per year. And so they currently do make up a significant fraction of our total calories, something like 5% globally. But if it can only grow a relatively small percent per year, you just can’t feed everyone on it.
Robert Wiblin: So it could stay at about 5% if we eat half of them [crosstalk 00:33:14].
Dave Denkenberger: Yeah. I think I estimated up to 10% after five years.
Robert Wiblin: So it just doesn’t scale?
Dave Denkenberger: Right.
Robert Wiblin: That’s mushrooms. What’s the next most promising option?
Dave Denkenberger: Well, you mentioned insects. There’ve been quite a bit of interest lately in insects as human food. Of course, many people around the world have eaten them traditionally, but not so much in Western cultures. They’ve been promoted for environmental reasons. They might convert food more efficiently. Though, if you’re concerned about their welfare, that might be a problem.
Dave Denkenberger: And there has been some work on feeding waste products to insects. But I haven’t seen feeding fiber, basically, like wood. And that would be important in one of these scenarios. But there certainly are insects that can eat wood like termites.
Robert Wiblin: Could we live on termites?
Dave Denkenberger: I think they’re pretty nutritious. Again, well, an animal will typically have a lot of protein and lipids or fats but won’t have the carbohydrate. There’s potential if you just have that as your food source, then it might be like being on the Atkins diet where you’re actually losing weight instead of gaining weight. So you need to have some carbohydrates-
Robert Wiblin: Excellent.
Dave Denkenberger: But lots of these other ones would have carbohydrates like the leaf extract and the mushrooms.
Robert Wiblin: How much mileage do you think we might be able to get from insects and rats and things like that, so setting aside any ethical issues?
Dave Denkenberger: Well, the rats one is a little more complicated because they are not as good of a cellulose digester as some of the other animals that we’ve talked about. And so the cue I took is actually from nature where fish will eat partially decomposed leaves, and fish cannot digest fiber. So why are they doing this? Well, they’re doing it to get at the bacteria that are growing on the leaves that the fish can digest.
Dave Denkenberger: And so my thought is we might be able to do the same thing to partially decomposed leaves and feed them to rats or even chickens which, like humans, have very little ability to digest cellulose.
Robert Wiblin: How much mileage do you think we might be able to get out of this category if we tried?
Dave Denkenberger: Well, I produced a graph of how fast we might be able to ramp up the different food sources. Some things could happen very fast like extracting food from leaves. You don’t have to wait for any organism to have offspring. Mushrooms can have a billion spores, so they can grow very rapidly.
Dave Denkenberger: The insects are somewhere in between the large mammals and the mushrooms, so they don’t ramp quite as fast. Chickens are actually quite fast because they can lay an egg a day, and rats are pretty fast too.
Robert Wiblin: So there is decent potential to scale there quickly if we’re organized about it. Let’s move on to things that I think we currently don’t really eat that much, which is growing bacteria with methane and breaking down plant matter with enzymes and bacteria. Let’s maybe do the enzymatic sugar one first. How would that work, and how much food could we get from that?
Dave Denkenberger: Well, we can look at an existing cellulosic biofuel plant and basically interrupt the process, at sugar. But what I’m interested in studying is there are, potentially, some toxins in that. The question is could then we purify that sugar such that people can eat it?
Dave Denkenberger: But once you do that, we eat sugar now. It turns out that there are different types of sugar and some cannot be digested by people, and so we would feed that to animals.
Robert Wiblin: What does that look like? Do we have to have just lots of bioreactors breaking down plant matter and then I guess processing it after that and feeding it to cows? How complicated is this?
Dave Denkenberger: Yeah, that’s right. Right now, we don’t have very many cellulosic biofuel plants. So the bigger question is could we retrofit other chemical plants to do a similar thing? And that’s one area I’d like to do, to work with a chemical engineer to say, “Well, how feasible would it be to retrofit existing plants to do this?”
Robert Wiblin: If you’re getting this broken down [human inedible] sugar coming out of these plants and then we’re feeding it to cows and sheep and so on, then we’re back at square one because you can’t get those animals to reproduce very quickly. But could you get other bacteria to digest it and then eat that bacteria or something like that?
Dave Denkenberger: Yeah.
Robert Wiblin: What are the pros and cons of that?
Dave Denkenberger: Depending on the bacteria, there can be issues with human edibility. But there are certain types of bacteria that are fine for people to eat, and I use the example of Spirulina which is known as a super food. People eat it as a supplement. It’s actually in some health drinks that look very green. It’s not the same type of bacteria. It’s actually cyanobacteria it’s photosynthesizing.
Dave Denkenberger: Many people call it algae, but it technically is bacteria. So people can eat that, and there are many examples of food that have certain amount of bacteria in them like yogurt.
Robert Wiblin: And so you just think that this technology already exists? Do we need many technical breakthroughs to make this practical?
Dave Denkenberger: There’s certainly a lot of research on how to make this cellulosic biofuels competitive with existing gasoline. But it turns out gasoline is relatively inexpensive compared to most food we eat. And in a catastrophe, we’d be willing to pay much more for food. So I think there’s a lot of potential for this to be cost-effective in a catastrophe.
Ever eaten any alternative food substitutes?
Robert Wiblin: Have you actually eaten any of the alternative food substitutes that you’re suggesting, especially the weird ones?
Dave Denkenberger: Yeah, one example is the spirulina. It’s very green and it does taste like spinach. Then, I’ve also eaten a number of different insects and there really isn’t very much of a strong flavor, so you mainly get the spice that’s been put on them.
Robert Wiblin: Well, let’s move on to the next one, which is growing bacteria with methane, which is something that I wouldn’t have guessed was a possible food source. Is that technically hard to do, and can we just then eat the bacteria that manage to grow on methane?
Dave Denkenberger: It is a pretty amazing thing that we have a life form that can not only use methane or natural gas as a food source, as an energy source, but also, it uses the carbon to build its own bodies. Now, you do have to add in some nutrients like nitrogen to make it work. But the basic process is relatively straightforward.
Dave Denkenberger: And there are a few companies around the world that are looking into using stranded natural gas, which means too far away to sell and to market, to produce the bacteria as fish food because it’s high protein.
Robert Wiblin: Is this already done anywhere?
Dave Denkenberger: I believe it has been done. I’m not sure if it’s commercially available. But yeah, they’ve done it at fairly large scale and put tens of millions of dollars into demonstrating pilot plants.
Robert Wiblin: Do you know where we got this bacteria? Are they from hydrothermal vents or deep under the ground where you don’t access to food that comes from sunlight?
Dave Denkenberger: I don’t know the history, but yeah, they are called extremophiles because they can live in an extreme environment, to be able to use methane as a food source.
Robert Wiblin: How much methane or how much natural gas would we need? And would it be conceivable that in a disaster, we could continue getting this much natural gas and spreading it around to everywhere that needs to produce food?
Dave Denkenberger: It would be a lot of natural gas. With the current production of natural gas, I think we could feed about half of the people. But again, we’d probably only be producing 10% of our calories from this or less. So it could be a significant food source. Then as for the catastrophe, again, it comes back to the question, “Is infrastructure still functioning?”
Dave Denkenberger: Then also, I’ve looked at how fast we might be able to scale it up. It takes a certain amount of energy to do your energy production infrastructure. But in the case of fossil fuels, it’s a pretty small percent. So we could technically ramp it up quite quickly.
Robert Wiblin: I guess again, because you’re using bacteria, they can just replicate incredibly quickly. They’ll have some pretty short doubling time.
Dave Denkenberger: It’s not limited by the bacteria. What I was referring to is what’s called embodied energy. So how much energy does it take to produce the drill that is drilling for natural gas?
Robert Wiblin: Oh, I see.
Dave Denkenberger: There’s some physical limit for how fast you could ramp that up. But as long as it’s only a few percent of the energy you produce, you could actually double in less than a year.
Robert Wiblin: With all of these solutions that involve bacteria, do you have an issue with preventing pathogenic bacteria or viruses in invading your vats and taking them over? Or I guess it’s sterility.
Dave Denkenberger: Well, in the case of methane, I don’t think too many random bacteria would be able to live in that environment. But in other cases like leaves and wood, yes, you do have to worry about that. And what we do in the case of mushrooms, we actually often grow them on manure. We first pasteurize the manure, and the cheap way of doing that is having a big pile of manure.
Dave Denkenberger: Then first, the bacteria start growing on it and increase the temperature a little bit. Then high temperature bacteria take over. So the low temperature bacteria actually die off, and then the high temperature bacteria keeps getting it higher and higher. Then it’s only the low temperature bacteria that can actually live in humans, or would be able to live later on.
Dave Denkenberger: And so you’re basically killing off those competing bacteria and then you grow mushrooms on them, and you need to keep it fairly well-sealed so that you don’t get re-contamination. But it’s a commercially mature process.
Robert Wiblin: Interesting. Do you have a problem with mushrooms that humans can’t eat also growing on the manure?
Dave Denkenberger: Well, presumably, they would be killed as well in that pasteurization process.
Robert Wiblin: I see, and then you only spread the spores of the mushroom species that you want.
Dave Denkenberger: That’s right.
Robert Wiblin: All right, so we’ve gone through a bunch of interesting ideas. One that I just remembered is that to my surprise, in a nuclear winter scenario or in a situation where the sun is largely blocked out, we’d actually be able to get more fish probably than we could now. Can you explain that?
Dave Denkenberger: Sure. The logic is that the vast majority of the oceans now are called an ecological desert, obviously, not because of lack of water but because of lack of nutrients. If there were a massive sun blocking scenario, the earth would cool and the upper layer of the ocean would cool and sink, and then deeper layers of the ocean would be brought to the surface, which have more nutrients.
Dave Denkenberger: And so we would have more nutrients available, but we have the disadvantage of less light available, and depending on the scenario, maybe more ultraviolet radiation. It needs more study, but I did do a scenario because I thought it was … in a way, it would be more feasible than relocating plants on earth because you don’t have the issue of different soil type and such.
Dave Denkenberger: Though, I would say in recent work, I have been looking more into whether we’d be able to relocate plants on the ground.
Robert Wiblin: How much would that increase the productivity of the oceans?
Dave Denkenberger: Well, right now, in the open ocean, the productivity is extremely low. Because of that low production, it’s very low density of production of photosynthesis and the phytoplankton, which is just small plants growing basically. Then in order to concentrate that to produce fish, it has to go through many trophic levels.
Dave Denkenberger: So you have the zooplankton that are eating the phytoplankton, and then you have … I don’t know. I’ve heard it can be up to seven different trophic levels. And so you start with low efficiency to start with, and then you go through all these trophic levels and you just produce hardly any fish. But in areas where there’s active upwelling of the ocean, the statistic I saw was .1% of the ocean area, so one thousandth of the total produce 50% of the global fish catch.
Dave Denkenberger: Because there, you have abundant nutrients and you can have a very short food chain. It can be fish that we can catch eating algae directly. So it’s way more efficient.
Robert Wiblin: If the amount of coast that has this updrafting increased a couple of fold, then the productivity of the oceans could increase two or threefold.
Dave Denkenberger: Yeah. It could be a really big increase. Now, there is the issue that once the climate stops cooling down, you’ll stop getting that overturning. Eventually, you consume the nutrients near the surface either because it’s people pulling the nutrients out in the form of fish, though, we might be able to put our waste back there, or material falling out into the deeper ocean.
Dave Denkenberger: So it would only work for a certain amount of time, but then it may be possible to continue that by … If agriculture is not working on land, we’re not using our fertilizers, we might actually just put the fertilizers into the ocean.
Robert Wiblin: Oh, wow. Wouldn’t it be more efficient to eat seaweed or algae directly rather than the fish that eat them, because you don’t lose the energy in the conversion?
Dave Denkenberger: Absolutely. But I have not been able to get the time to run that scenario. But potentially, you could feed a lot more people that way.
Summary of the overall picture
Robert Wiblin: Do you want to summarize the overall picture? You’ve had all of these interesting ideas about how we might feed people that mostly others haven’t investigated. Collectively, how optimistic are you that we’d be able to keep everyone alive?
Dave Denkenberger: From a technical perspective, what we could do … I am quite optimistic because even though some of these solutions might not work out as well as I think they might, we do have quite a bit of redundancy in the system, that is, when I analyzed the food sources individually, many of them could increase up to feeding everyone fairly quickly even in one year.
Dave Denkenberger: Now, in reality, they would be competing for energy sources, so it’s not quite as good. And I did write another paper which actually analyzed them including the interactions but still found that we could feed everyone two times over or three times over. That’s what’s technically possible.
Dave Denkenberger: The other thing is that I’ve come up with another, and some people have given me more ideas based on feedback, that there are other things that just haven’t analyzed yet. One of those is seaweed, like you said, but other things like using energy from fossil fuels to directly synthesize food.
Dave Denkenberger: We’ve already done that at the lab scale. So the question is how fast could we ramp it up? I just haven’t done that analysis yet. Or bacteria that run on electricity. We’ve talked about how we have methane-eating bacteria, but what about nuclear energy? Well, that’s electricity. Well, maybe that could actually grow bacteria, and there are examples of electric bacteria.
Dave Denkenberger: By the way, I would say one other solution that is commonly proposed by people when we talk about this issue is well, why don’t you just grow plants indoors on your electricity? But it turns out it’s extremely inefficient to take electricity and turn it into light, and then do photosynthesis. Even if you have the most efficient algae using all of our electricity, it would only feed, I believe, 5% of the people. So it’s way, way better to use the natural gas directly on natural gas digesting bacteria.
Robert Wiblin: Where are we losing the energy there?
Dave Denkenberger: The biggest loss is actually photosynthesis. So you produce the light and the typical crops in nature might be .3% efficient. If you’re really good, maybe 1%. And then algae, under ideal circumstances, maybe 3%.
Robert Wiblin: So photosynthesis just isn’t that effective?
Dave Denkenberger: Right.
Robert Wiblin: This raises the issue, I imagine that almost none of our calories from any of these methods now. Basically, all of our food comes from the sun fairly directly. Would it be sensible to switch our food supply now so that more of it is produced using these other methods even just because it might be cheaper or these would be just good ways of producing calories for people?
Dave Denkenberger: Yeah, potentially. Of course, you need to make it cheap to compete with current food. But I do think that thinking more about how we can utilize waste … I mentioned how cows used to eat corn stalks. But now in developed countries, agriculture has been disintegrated. That is, we keep our cows separate from our plants and we don’t have that traditional integrated farm.
Dave Denkenberger: We could potentially either reintegrate or ship the corn stocks to the cows and reduce our environmental impact. Even if it’s not cost-effective right now, people may pay a premium just to reduce their environmental impact. Another example is when we log forests. There’s a huge amount of material, woody material that’s left over. It can be higher weight than the actual lumber we take out. But mushrooms can grow on that. I mean why don’t we get some food out of it?
Robert Wiblin: Do you think that, setting aside any disasters, any of these methods would be applied just to feed people for normal reasons within the next 50 or 100 years?
Dave Denkenberger: I think potentially, it all depends on the economics. As you pointed out, if we do have difficulty with conventional food because of, say, slow climate change, population growth, many other problems like loss of soil, soils getting salty, any of these things, they’re not quite the big disasters that I’m working on, but they can add up.
Dave Denkenberger: And if conventional food price increases enough, then I think these other options could come in. The other possibility is local disasters. Generally, if you can just ship outside grain in, that’s going to be the cheapest. But maybe transportation is cut off, and so you might want to know how to grind up your wheat leaves and get some more food out of them.
Robert Wiblin: It surprises me that we haven’t done this before. I guess it’s just that food is currently just so abundant that we don’t have to think very creatively of other ways to get calories.
Dave Denkenberger: And even just in the last 50 years, the long-term trend has been reduction in inflation-adjusted food price. So it’s not thought about too much.
Robert Wiblin: You wrote this book where you discussed all of those options back in 2014. Have you changed your opinion since then about which one of these are most promising and which ones you would like to push the hardest?
Dave Denkenberger: Well, I have done some initial estimates of how much it might cost. Of course, it’s a complicated question for what they might cost in a catastrophe. But I’ve just said, “Well, how much do they cost now?” to give us some idea. The lower cost ones included the natural gas digesting bacteria, the sugar production with enzymes and leaf extract, potentially, low cost fish.
Dave Denkenberger: I didn’t mention that the fish we’d probably be eating in a catastrophe are very small, ones that could algae directly like sardines, and that also breed very frequently. But then another one, as I alluded to, is the potential of relocating crops. That’s something I didn’t include in the book. But I’m actually setting up an experiment now to simulate the conditions of nuclear winter in the tropics to see if plants can grow.
Dave Denkenberger: I think it is possible. Of course, there’s uncertainty, but I’m basing it on the simulation where about half of the sun is blocked in about an eight degree Celsius reduction. And so we might be able to grow potatoes, for instance.
Robert Wiblin: How much food do you think we would be able to get from that?
Dave Denkenberger: Well, that’s what I’m going to try to estimate based on this experiment. But even if we can get a quarter of our food from that, then it makes it easier to produce the remainder-
Robert Wiblin: The remainder.
Dave Denkenberger: Of these alternate foods.
Robert Wiblin: It sounds like with a lot of these solutions that you have suggested, there isn’t a lot of existing research. So you’re having to guess quite often, and I think you’re only really claiming, in many of these cases, to be right to the nearest order of magnitude. So it could be three times worse or three times more effective than what you’re suggesting.
Robert Wiblin: What are the pros and cons of having just such a rough view of all of this? I suppose it means you can cover a lot more ground a lot more quickly. But on the other hand, it’s a bit difficult to compare what should be your greatest priorities because there’s just not a lot to go on.
Dave Denkenberger: I think that that’s basically where you have to start. As you say, we don’t have very much data to go by, so there has to be a lot of estimation. There’s some things where we can’t say anything useful, but other places, we can. We have some idea even though there’s uncertainty in how fast we could ramp up different food sources, we can still say mushrooms will ramp a lot faster than cows will.
Dave Denkenberger: I would say also, in the cost-effectiveness, in some cases, you just have to say, “Well, it’s similar cost-effectiveness to something else. But other things, you can actually have reasonable confidence that it’s more cost-effective.”
Robert Wiblin: How did people react to this kind of first cut? Because I guess it seems very good to me that you’ve done this … You’ve put the first brick in the wall here, and you’ve said, “Well, here are the main questions that we need to answer within this field, and no one else has really been working on it.”
Robert Wiblin: But I know sometimes academics don’t really like doing things that … They want to do one thing very precisely rather than cover a lot of ground vaguely. Have you had a negative reaction to the fact that you’re trying to do so much without a lot of existing literature to work with?
Dave Denkenberger: It certainly has been a challenge going through peer review. Many people don’t feel qualified to review it because it covers so many different fields. When I’m writing it, I always get experts to review like a mushroom expert or a rabbit expert. But you can’t have 10 different reviewers on a paper.
Dave Denkenberger: So it certainly has been a challenge. But eventually, we’ve gotten something like seven peer reviewed papers so far, so it is possible. Generally, the way I construct uncertainties like a distribution, is that wide enough such that most people will say, “Yeah. My value is in there somewhere,” and then generally, people are happy.
Dave Denkenberger: Sometimes I’ve been criticized for having too wide of distributions. But I think that’s certainly one of the lessons I’ve learned from the rationality community, is that you should have really wide distributions. And usually, people understand that that’s appropriate.
Robert Wiblin: I’ll just add a note before we go on to the next section, that regular listeners will know that I’m vegetarian and here, we’ve been talking about eating animals. I guess we haven’t even considered the ethical issues here at all. Although, I imagine the ethical calculus would be a bit different if starvation was the alternative.
Robert Wiblin: Though, it does also sound like animals, in most cases, is not going to be the most efficient way to produce food even in a disaster scenario. But I just wanted to bracket that so people don’t email me. Let’s move on to envisaging the world that this would look like. I guess we can have a more concrete idea of actually … to think through how useful would it be to try to prepare to do these things sooner ahead of time.
Robert Wiblin: You said earlier that you thought only some of the world’s infrastructure would be destroyed and only a small fraction of the world’s population would be dead in a nuclear war. Do you mind painting out in any more detail what infrastructure you think would be destroyed and what would still remain? How many people would be injured, say, in this disaster so that we can just try to visualize what a city would look like in this case and what people would be doing day to day?
Dave Denkenberger: Again, there’s uncertainty, and I did actually get into this somewhat in the paper that looked at nuclear winter uncertainty, that basically, if you have the really bad scenario both from killing a lot of people and from the nuclear winter scenario of targeting the population centers, there would still be … Depending on how far away you are from the bomb, there is a criterion of how much pressure the blast wave is, like roughly what the percent of survival is that we’ve gotten from the Japan experience.
Dave Denkenberger: But yes, you could have injuries and then obviously, any healthcare would be overwhelmed, so you’d basically be on your own. But the mortality would not be 100% even in the cities. In the rural areas, yes, there would be radioactive contamination and increased cancer deaths. But it would not kill the majority of people in the rural areas.
Dave Denkenberger: The situation is even less extreme in non-target countries, though many people think, and there’ve been popular movies and such that, “Oh, if we detonated these many nuclear weapons, then everyone in the whole world will die from radiation poisoning.” But that’s not actually the case because the radioactivity generally rains out. Some of it right away but most of it, within a few days.
Dave Denkenberger: Of course, some would travel up into the stratosphere that could stay for years. But at that point, if it’s then only raining out over years, the exposure is just much lower. That’s why we’ve focused on what is the biggest problem, which is the food.
Robert Wiblin: I guess I’ve somewhat ended up, I think, with a distorted picture of what a nuclear apocalypse would look like, because I watched this movie, Threads, which is from the 80s which followed a city in Britain during an all out nuclear war between NATO and Russia. In that case, I think a very large fraction of the population was dead, and most of the remainder died in the coming years of starvation or cold or-
Robert Wiblin: Died in the coming years of starvation of starvation or cold, or just their injuries. And basically all of the infrastructure was down, and basically the government fell apart. And I think that they were trying to be extremely accurate for what would happen in Britain in a situation. But I guess that is a worst-case scenario, because there’ll be so many nuclear weapons at that time targeted at a very small country that’s very densely populated. So the radiation is much worse and the number of cities that are targeted, and the number of people who are hit directly by nuclear weapons would be much worse.
Robert Wiblin: And so I guess you’re saying that I should imagine places where basically almost everything is still functioning except perhaps there is a bit like an uptick in cancer, and trade is, to a significant extent, cut off. But otherwise, things are still running.
Dave Denkenberger: Yeah, yeah, it depends on how we react. The basic assumption in the book of Feeding Everyone No Matter What, assumes that we continue to cooperate. Which means trade of goods, sharing information, etc.
Dave Denkenberger: But I have done some less optimistic scenarios, say, what might be an economic scenario. So you would still have trade, you would not have immigration, seeing just how much trouble we’re having with refugees at this point, that’s probably not gonna be a feasible solution in a disaster. But if you still have trade of goods and sharing of information, and then a world food price, I was able to estimate well, what percent of the population would survive. And if you just have for stored food, it’s only around 10% of the population. I estimate, even now without anymore research and development, if countries just knew about these solutions or were told in time before they resorted to further military action, we could do much better than stored food, maybe 30% or so of people would survive.
Dave Denkenberger: But if we actually got prepared, like some of these alternate foods need more research. Some have already been developed commercially, but we’d need to figure out how to scale it up quickly, say retrofitting factories. And we actually have plans for scale up, and plans for how we would continue trading and things like that. Then survival could easily be 60, 70, 80%.
Dave Denkenberger: So food is gonna cost more if there isn’t charity. But we could still have a much much better scenario than just stored food.
Robert Wiblin: So we’ve mostly talked here about the nuclear apocalypse situation. And I think that makes sense, because if we can deal with that, then that’s the worst case, and we should be able to deal with most asteroids, and super volcanoes, and more modest wars. Is that right?
Dave Denkenberger: That’s right. It is the worst case, especially because it’s longer lived. The black smoke particles from burning of cities would be heated by the sun, and actually lifted up higher, so they would stay there longer. Whereas the super-volcanic eruption, the particles are whiter, and the sun might only be blocked for a couple years.
Robert Wiblin: Are there any other interesting aspects of the other catastrophe scenarios that we should pay attention to?
Dave Denkenberger: Well, certainly some of them don’t involve blocking of the sun. And then you for instance wouldn’t have the overturning of the ocean. You could get a really large asteroid that can completely block the sun, in which case we’re not gonna be growing any potatoes or any seaweed. But generally, the solutions are applicable across a wide variety of catastrophes.
Robert Wiblin: So would the Internet survive this? ‘Cause I was thinking … I mean a very important aspect of the preparation would be that you have to share this information with everyone around the world, so that they know very quickly what they should be doing. But if the Internet’s down, it’s much harder to communicate, then that’s something that you need to prepare for ahead of time.
Dave Denkenberger: Yes, that is extremely important. But fortunately, the Internet was actually designed to survive a nuclear war-
Robert Wiblin: Oh, wow, okay.
Dave Denkenberger: … back in the 60s or so.
Robert Wiblin: So everything can just be rerouted through different parts.
Dave Denkenberger: That was the idea, yeah. And so that was even trying to keep connectivity within the US, and of course it’s much easier in the non-affected countries or non-target countries.
Robert Wiblin: What about other issues like people dying because it’s really cold, or they don’t have access to water, have you considered those, or is that just someone else’s job?
Dave Denkenberger: I did do a quick survey in the book of the different problems. My background is actually in energy. So I was interested in that part of it. So there’s a question; can the buildings be heated adequately? There may be some circumstances where we can’t do that and we need to relocate some people, or people would do makeshift stoves and burn wood. So it seems to be not too hard.
Dave Denkenberger: Now, the water issue is particularly interesting, because if the earth cools, you get less evaporation from the ocean, and that ends up in less precipitation on the land, something like only half as much, which sounds really bad. But it turns out, more than half of our water is actually used for growing food. So if we’re not growing food, we could use the water for other things.
Robert Wiblin: Okay. So I guess drinking water is the last thing that you would stop using. So if you cut back on showers, then you’re not going to die of thirst.
Dave Denkenberger: Yeah, and it turns out the major uses of water are agriculture and cooling power plants, even showers is relatively small, and the actual drinking water is minuscule compared to those other things.
Robert Wiblin: All right. I guess that it’s the first benefit of most agriculture being gone.
Dave Denkenberger: Yeah, yes.
Robert Wiblin: So a critique that I’ve seen in some comments on your work is that in this scenario, with a nuclear winter or an asteroid, once that happens, kind of so many people’s attention is gonna be turned to how do we continue feeding ourselves, ’cause that’s just going to be obviously a huge potential threat to society? And given that you might then have hundreds of millions of people, thinking about this question after the fact, how much value do we really get from preparing, having a few dozen people prepare for it ahead of time?
Dave Denkenberger: Yeah, it’s a good question. There are some historical analogies, the siege of Leningrad, I believe, was a case where the city was cut off for years, and many people just starved. I mean you could argue now that we have better technology, and better educated people, so we should have a better chance. Certainly if we do have free flow of information, that makes it easier, they might find the work that ALLFED has done so far just on the Internet, or yeah, people could invent it independently. So there’s certainly that possibility.
Dave Denkenberger: One concern I have is just that leaders react quickly before the information gets out, and chose the military route, and then you can easily have a downward spiral.
Dave Denkenberger: The other thing is that … I mean I do think that further dissemination of the research we’ve done so far would be beneficial. But we would certainly have a better chance of success if we did more work, like studying long-lived animals, you can’t do that overnight. Or figuring out what plants might actually grow in the tropics. Well, it takes a long time to grow plants. And we can do things in the greenhouse, but then we have to wait three months. So it would be hugely advantageous to actually have figured that out ahead of time. And then have a plan, so that we could start implementing it right away. Because every delay you have, means fewer people are gonna survive.
Robert Wiblin: Okay, so just the first one is that you think it’s very important that people be made aware very quickly that it will be possible to feed everyone, and so they don’t need to go out and fend for themselves, and we see kind of a breakdown of the legal system, or of security.
Dave Denkenberger: Right. Either within countries or between countries.
Robert Wiblin: Yeah, how confident are you that that’s a big effect that for example if we promoted this message, that we would be able to persuade people of that, and that it would make the difference between the legal system continuing and kind of a bigger breakdown of security?
Dave Denkenberger: Not very confident, and basically we’ll talk about later about the actual cost effectiveness modeling. But there’s some probability that it will work out well based on our current preparation. If we prepare more, that probability increases. And you could certainly imagine spending trillions of dollars, like storing up food, and then we would be extremely confident that it would work, because we wouldn’t have to have any new technology work. We would just have the stored food.
Robert Wiblin: So I suppose it then becomes important to be able to persuade people that these technologies will work, which I guess is another reason why you wanna be able to demonstrate them on a smaller scale first.
Dave Denkenberger: That’s right. So that’s one of our priorities now, is to demonstrate technologies on a small scale. And it has a number of advantages. One of course, doing a small-scale experiment is less expensive. But then the other thing is that depending on the scenario, if we do have breakdown in cooperation, then we may not be cooperating well enough to retrofit factories. But if we could have something that grows food in people’s basements, then that could still save more lives. And I’ve been focusing on saving lives, but the primary motivation for me is the long-term impact. And so I just think there’s a very strong correlation between saving lives and actually retaining a functioning civilization.
Robert Wiblin: And you’re saying that, I guess, the thing that can’t be done after the fact very quickly is just many rounds of testing to figure out how to do this, right?
Dave Denkenberger: Right.
Robert Wiblin: Because you can’t parallelize that, if you have a 100 million people doing it, it takes a month to test it out, and then you gotta go back to the drawing board and rework the prototype. So that, you have to do ahead of time.
Dave Denkenberger: Mm-hmm (affirmative).
Keeping the peace in GCR scenarios
Robert Wiblin: In terms of us being able to keep the peace in a disaster scenario, so long as the government tells people that it has a plan for feeding everyone and so they need not turn to violence, do you think that people would believe the government if they said that? Is there kind of going to be an issue of credibility that people might think, “Well, of course, they’re saying that. They just like want to trick us into going along with them.”
Dave Denkenberger: Yeah that was something that came out of EA Global San Francisco Workshop I did in 2016, where the general thought was that of course the government is going to say, they have a plan, but unless they could credibly back that up with what they had done ahead of time, then many people would not believe them. I think that’s a good reason to do at least some planning ahead of time.
Dave’s additional point about post GCR scenarios
Dave Denkenberger: Another concern I have is that in the immediate response to the catastrophe, there may be lots of suggestions of ways of handling the catastrophe. The best ones technically might not win out, depending on the charisma of the advocates. One example we give is the dependence on using all our fossil fuel energy or electricity to make artificial light and grow plants that way. Because that’s an extremely inefficient way of producing food, we could use literally all our energy for that and not feed very many people. The fact that we’ve tried to think ahead about how we could allocate resources efficiently, I think would help in the catastrophe time when not everyone will probably be thinking rationally.
Robert Wiblin: Okay. So what is ALLFED’s plan? If it had a million dollars, or $10 million or $100 million, what kinds of levers would you be pushing?
Dave Denkenberger: Well, I see it as a two-pronged strategy. One, as I’ve said, I think that greater awareness of what we’ve done so far would make it more likely that we have a good outcome. But of course it’s expensive to get the message out [ahead of time]. But we could be prepared to get it out very quickly.
Dave Denkenberger: So one strategy would be to have a panel of experts, and this could actually apply to even beyond agricultural catastrophes. But have people trained in the media, and aware of how to react, and then say to the media, “Well, we don’t need a lot of your time now, but if some of this happens, call us and we’ll actually have a message that’s not just everything’s going to hell and panic.” Which the media tend to like to say.
Dave Denkenberger: Similarly, maybe instead of the mass media, it could be on social media. It’s very difficult to figure out what goes viral now. But if people are of the sudden interested in feeding themselves, we could be prepared with a message that could potentially go viral at that time.
Dave Denkenberger: And then a third way is to find influential people who it would be very advantageous for them to know ahead of time. And one example I give is Tim Benton who was the UK government food security champion. And even though he’s not the Prime Minister, that’s his [Benton’s] job of food security, so you would think he might be able to get the message [to the Prime Minister] in a catastrophe.
Robert Wiblin: Okay, so this kind of the advocacy and message spreading thing, is there also … I mean how much infrastructure should be built ahead of time? Should we have appropriate mushrooms spread out all over the place so that we can very quickly start growing them?
Dave Denkenberger: Yeah, so I guess I’m just saying the information sharing, why that’s the first prong is one of these catastrophes could happen this year. And so I think we should spend some of our efforts to say what would we do if that happened. But of course then we should further develop the technologies to have a better message to spread if the catastrophe happened later.
Dave Denkenberger: As for the preparation beyond the research, development and planning, there are many solutions that we could do that would cost a lot of money, other than food storage. We could, ahead of time, retrofit plants, or build new plants, chemical plants, so that they could be very easily switched. That would be great, but it might cost billions of dollars. So I wouldn’t do it first.
Dave Denkenberger: So basically there’s a big supply curve, basically, of risk mitigation. And what I wanna identify is the ones that could give us the biggest bang for the buck. That’s what we should do first.
Robert Wiblin: Is ALLFED doing very much media at the moment in order to kind of spread the word about how we could feed everyone in a disaster, so that people will stay more calm after the fact?
Dave Denkenberger: We have done some of that though we have been updating recently based on EA feedback on the potential downsides for mass media outreach now. Our thought more now is to be ready for a catastrophe that is have relationships with journalists, so that they would contact us in a catastrophe and then, hopefully, we would get this more positive, cooperative information out rather than the typical doom and gloom in a catastrophe.
Robert Wiblin: Yes, do you need to build up an experience with media? I suppose you’re doing that now to some extent.
Dave Denkenberger: We have had some experience with interviewing, though we do want to do more training. Ideally that we work with other organizations, so that they could react well in a catastrophe. It doesn’t have to be necessarily just agricultural catastrophes.
Robert Wiblin: Earlier you mentioned doing an experiment to figure out how much food you could grow in the tropics in a nuclear winter scenario. Are there any other experiments or trials that you’re thinking of running?
Dave Denkenberger: Yeah, lots of them. So I’ve talked about the methane-digesting bacteria being done at commercial scale, it’s potentially possible even at household scale, assuming natural gas still flows. So I’d like to try that out, see how efficient it might be.
Dave Denkenberger: It’s also been suggested that … UK listeners might be familiar with Quorn, with a Q, which is fungus-based protein source, often as a meat substitute. And that is grown more like industrial, like in vats. So that might be a cheaper way of converting cellulose into food than growing mushrooms. So that would be great to investigate small and large scale.
Robert Wiblin: Yeah, that makes me think. It sounded like some of the options you’re considering could have commercial applications today, and potentially even if it’s not exactly what you would do to prepare for a disaster, if you could build a business that figures out a way to grow mushrooms on wood in a way that’s profitable now, then it could be at a much larger scale, should one of these disasters happen. You wouldn’t need donations, ’cause you could fund it just by selling the product.
Dave Denkenberger: Yes, and we’re certainly open to that. And I’ve heard one critic say, “Well, but then if it becomes mainstream, then you don’t have the ability to ramp it up quickly.” But that would only if it were very large. So still I think it would be of benefit to commercially develop more of these technologies.
Robert Wiblin: Sorry, I don’t understand that. Let’s say that we were getting 50% of our food through Quorn hypothetically, this vat-grown mushroom stuff, then doesn’t that just mean that we’re not so dependent on sunlight anymore, even if we can’t scale it up that much more?
Dave Denkenberger: Yeah, I guess that is true. You still need to have a supply of the energy source. But yeah, if it’s using wood, then we would have lots of wood.
Robert Wiblin: Okay. So let’s switch now to talking about the cost effectiveness analysis that you’ve recently got published. And actually, it looks like you’ve got two different papers published; one cost effectiveness for interventions for alternative food to address global catastrophes globally. And the another one doing, I guess, a similar analysis just for the United States. Do you wanna describe how you did that cost effectiveness analysis, and what you found?
Dave Denkenberger: Sure. So in the case of the United States, it gets back to what I talked about before, the modeling of nuclear winter, because really that’s the only catastrophe that could cause starvation in the United States. There of course could be other bad consequences, even if there’s not starvation in the United States, like a refugee crisis, or a conflict abroad that spreads. But the easiest model is starvation in the US. And for that, the cost effectiveness turns out to be very large, even though it’s only just from one country selfishly to prepare for these alternate foods.
Dave Denkenberger: And it’s partly because the alternatives for saving American lives are extremely expensive. Like healthcare or environmental regulation or traffic safety.
Dave Denkenberger: But what is probably more interesting to your listeners is looking at the global case. And for that, I actually did the 10% shortfalls that we’ve been talking about. Because those can cause mass starvation globally, as I said, not because there is technically not enough food, but because of price increases and the poor of the world not being able to afford it. And as I said, there is tremendous uncertainty in what our response would be, the mortality could be as low as millions, or it could even be billions if it goes very poorly. But the expected value is in the hundreds of millions if one of these catastrophes happened.
Dave Denkenberger: And in this case, we’re comparing to actual GiveWell estimates of saving lives of saving lives with mosquito bed nets. And that, the cost to save a life is in the thousands of dollars.
Dave Denkenberger: As you mentioned earlier, my analysis indicated it was more like tens of cents to hundreds of dollars to save an expected life. And this is not … Assuming the catastrophe will definitely happen, this takes into account some probability that the catastrophe will not happen, and actually only looks out a couple decades, because things could change dramatically. For people in a general audience, the research we do now could get out of date, or planning we do now could get out of date. If you’re worried about risk from artificial intelligence, then something much more dramatic could occur, and then these things would no longer be relevant.
Robert Wiblin: So do wanna kind of map out what the key parameters were? And one thing that I really like about the cost effectiveness analysis that you’ve done is that you’ve put it into this Internet talk or Guesstimate, where you map out ranges or probability distributions for all of the different parameters that are feeding into the ultimate life saved per dollar figure. Which makes it a lot easier to kinda scrutinize and figure out where you agree or disagree.
Robert Wiblin: But maybe what are kinda the key numbers that you had to estimate to figure out how many lives you might save per dollar?
Dave Denkenberger: Well, in this case, these papers were actually written before Guesstimate existed, so I was using the software Analytica, but it’s still possible that you could view it for free and modify assumptions if you’re interested in doing that.
Dave Denkenberger: But there are a number of assumptions that go into this, we’ve already talked about some of them. But some other important variables are … In this case we’re talking about saving lives, so the probability of feeding everyone if we don’t do anything more, versus if we do these interventions, and I actually broke out three different, actually four different interventions. One was this planning, one was more research, especially on the technologies that have not been researched very much. One was on development to actually commercially develop the more promising technologies.
Dave Denkenberger: Those all cost something around $100 million. I also consider a more expensive options which is training. So here, you can easily spend billions of dollars, because it involves lots of people, and it’s actually maybe even periodically or every year or something, running drills. Obviously the military is very good at this. So then they know exactly … They are fully trained, they know how to respond. That would provide more value. It would increase the probability of success, but it would cost billions of dollars. And so it’s not the first thing I would do.
Dave Denkenberger: So then you put all this together in a Monte Carlo model, which a probabilistic model to give you outputs, and they are, as we said, they vary orders of magnitude. But in the case of cost per life saved, even though the variation was four orders of magnitude, it still really didn’t overlap with the estimates of saving lives with mosquito bed nets. So you can actually say with some confidence, if you believe in the inputs, that this would be better.
Robert Wiblin: So the high end in this paper, it looks like it was $400 per life saved, which I guess compares pretty favorably to the few thousand dollars that give all things that takes to save a life with bed nets.
Dave Denkenberger: That’s right. That’s for the three relatively inexpensive interventions.
Robert Wiblin: So in this paper, you’re estimating just a very low cost per life saved, do you wanna kind of justify? Is there any way of kind of intuitively justifying that to a listener who might be skeptical that that’s plausible?
Dave Denkenberger: Yeah, I would say that I think that there are clear things that we can do to increase the probability of feeding everyone. I think we would, as we said, be much better prepared with a plan, and with more research and development.
Dave Denkenberger: And then of course, you have to believe there is a significant chance of one of these catastrophes happening, which I think there is reasonable evidence for. But then the big picture is just that no one has worked on it before. So you could say that unintentionally, people have developed this methane-digesting bacteria, which is great for us that’s already been commercially developed. But they haven’t thought about how it would be done in a disaster. So we can just be much better off if we actually think that through.
Dave Denkenberger: And then there are other things that just haven’t been researched. And so I think we’re just at the very early stage where we can be highly cost-effective. We can pick that low-hanging fruit.
Robert Wiblin: All right, so in addition to that paper, you’ve also done an analysis where you looked at how likely this is to make a difference to kind of the long-term future of humanity, and lowering existential risk, do you wanna kind of talk about what’s different about that model? What gets added?
Dave Denkenberger: Sure. So we still have the probability of a catastrophe happening, and so we’re using the probability of nuclear war for one part of the model, for the sun-blocking catastrophes. But now, here, we’re not asking how many people would die, but we’re asking what is the impact on the far future? And my initial work, actually on a EA forum post, was considering one of impacting the far future, and that was a collapse of civilization, and then not recovering that civilization.
Dave Denkenberger: And so by some definitions you would say, “Oh, that’s not an existential risk, because people have not gone extinct.” But if you actually look back at Nick Bostrom’s original paper, there are several different types of existential risk. And he defines existential risk as something that prevents humanity from achieving its potential. And so if we lose civilization, and we continue on earth as hunter-gatherers with a few million population, maybe we would go extinct, because of the next asteroid. But even if we didn’t, we certainly are not achieving humanity’s potential. And so that does qualify, by that definition, as one existential risk.
Dave Denkenberger: Then after that work, partly based on feedback on it and also based on my reading of 80,000 Hours work, I shifted to a different perspective, which is well, there are many routes how these catastrophes could reduce our long-term potential. It doesn’t have to be just losing civilization and not recovering.
Dave Denkenberger: So one example this catastrophe if it goes poorly, many people die, lots of conflict, that could be a scar on the human psyche, and we could become nastier now. And I would say that would make it more likely that that nastiness gets put into artificial intelligence that we develop that then may determine our long-term future. And I think there are several other possible routes to having our long-term impact from these catastrophes.
Robert Wiblin: So it’s kind of there’s two different flows of value here. One is thinking about just like lives saves in a more normal sense, and then you are thinking about what’s the likelihood of it reducing an existential catastrophe which could put humanity on a worse trajectory, such that we never manage to achieve our full potential, whatever that is.
Dave Denkenberger: Right.
Robert Wiblin: There’s two different kinda cost effectiveness analyses that you could do. And I suppose there’s also two different scenarios. There’s the like an all-out sun is massively blocked case, and then also thinking about the 10% reduction in agriculture scenario. Out of those two, which one is generating the most value in the analysis?
Dave Denkenberger: Perhaps surprisingly, they are similar. And of course I say similar because that means the very large uncertainties are overlapping each other. So the 10% shortfalls are more likely to happen. But of course if they do happen, it would be less likely to have a far-future impact. So for this particular work, implemented on Guesstimate, which was developed by an EA, and it’s great that it’s very easy, like basically spreadsheet-based, that you can modify the assumptions if you disagree with them.
Dave Denkenberger: But what I try to do instead of just coming up with the distributions myself, I thought I would survey other existential risk research to try to get estimates of what is the far future impact likely to happen with current preparation, and how much would spending $100 million actually improve the outcome, both for the sun-blocking scenarios and the 10% shortfalls.
Robert Wiblin: So it seemed to me looking over this guesstimate sheet, I mean I wasn’t able to fully understand it, it’s pretty scrutable for a Monte Carlo simulation, but it does take some thinking through. And I definitely encourage people to go and take a look at it, ’cause it’s fascinating. And I think that it’s an approach to modeling that should be used a much more.
Robert Wiblin: So there was kind of four key numbers that you needed to estimate. One is the likelihood of a big food shortfall. Then there’s kind of the likelihood that that flows on and wrecks the long-term future, and then there’s the likelihood that kind of preparations that ALLFED might do would make the difference between it being really bad for long-term future, versus we recover from it and things go okay.
Robert Wiblin: And then I suppose you had to estimate how much that would cost. And if you can multiply those three, then divide by the cost, then that gives an idea of the benefit to cost ratio. Is that kind of right?
Dave Denkenberger: That’s right.
Robert Wiblin: So maybe let’s go through each of these in turn. So it looks like you estimated, or you had like a median of 1.9 [percent] risk of a full-scale nuclear war each year. Where did you get that number from? ‘Cause it seems quite a bit higher than figures I’ve heard from other sources.
Dave Denkenberger: Right, so I should clarify that in Guesstimate, the single value that is reported is the mean not the median.
Robert Wiblin: Oh, okay.
Dave Denkenberger: And this is a huge difference in the case very large distributions. But the background is that two of my colleagues at the Global Catastrophic Risk Institute, Seth Baum and Tony Barrett, looked back at the close calls for nuclear war. And on this podcast, you’ve gone into some of those. And many of them are very scary how close we might have come to nuclear war. And they built what’s called a fault tree analysis where you have to go through a certain number of steps for it to actually turn into nuclear war. So then they were able to actually put a quantitative estimate on nuclear war. And it’s very large range. The 95% confidence interval is something like 0.1% to 10% per year. But because of that large range, partly, but the mean turns out to be around 1 or 2%.
Dave Denkenberger: And that does sound pretty high, that is higher than most people’s estimates. And from one perspective, you should say that maybe we should update based on the fact that we have not had a nuclear war in the last 72 years. However, you could say we have had a nuclear war in the last 73 years. Now, of course the circumstance in World War II was different than now, but if you were to say that, then the annual probability would about 1.4%, which is in the ballpark.
Dave Denkenberger: Now, because it’s not the same, I do think we should be doing some updating downward based on the evidence. However, this model actually only considered inadvertent nuclear war, which means one side thinks they’re being attacked, and therefore quotes retaliates, and then the expectation is that there would be an actual retaliation afterwards.
Dave Denkenberger: But there are other scenarios that could lead to full-scale nuclear war such as an actual intentional attack, or there could be an accidental detonation of a nuclear weapon that could be misinterpreted and then escalate. So from that perspective, you could actually argue the probability should be even higher.
Robert Wiblin: Yeah, I think I still think that this number is too high. So what are some of the reasons? I guess it doesn’t seem right to say that we had kind of one … It’s true that nuclear weapons have been used once in a war in the last 73 years, but the situation was so different, just to use it over a single city, and the risk of that precipitating an all-out nuclear were so … Well, I mean obviously it can happen, ’cause they’re the only ones with nuclear weapons. That doesn’t really seem like it’s in the same reference class, ’cause everyone knows that if they used the nuclear weapons now, it would kind of be the end of their country in effect. And that wasn’t the case when it was used against Japan in World War II.
Robert Wiblin: So if you do the calculation where you say you had a 1.8% chance of a nuclear war per year, then over 70 years there should be a 72% chance of an all-out nuclear war over that period. So it seems like we get like a pretty big update against that having been the probability at least in the past, right?
Robert Wiblin: And in addition, that paper with Seth Baum where they did that [inaudible 01:26:45] analysis, it’s got … So the 90% confidence interval is between, I guess, 0.33% per year and 16% per year, which is … I mean ’cause that’s over like three or four orders of magnitude. It seems that they’re basically saying that they just have no idea. And then just to take kind of mean of people, giving this such a wide range.
Robert Wiblin: Yeah and then just to take the mean of people, given there’s such a wide range. It just seems like a lot of weight to put on something that’s basically saying we have no clue. What do you think of those ideas?
Dave Denkenberger: Well, I think that it’s good that it’s actually quantitative because so much of people’s guesses are not based on an actual quantitative model. So I would trust it more because of that reason. I would also say that we can’t just multiply the probabilities using the mean because there’s … And actually, I do this at the bottom of the guesstimate model, that if this were true, the 1.8% per year, what’s the probability remaining in the 21st century? If you use the means, it’s very high when you actually multiply out the distributions, which is what the accurate way of doing it, then you get 38% chance, but it’s still high.
Robert Wiblin: Yeah, okay. So what does that come to if you just had to pick one number then? I guess that’s like .5% chance a year, .6% chance a year?
Dave Denkenberger: Yeah, around there.
Robert Wiblin: Okay, so that’s less far off the estimates people give. I think Anders Sandberg wrote a paper where he said it was closer to .1%. Maybe that seems a bit on the low side because we have model uncertainty.
Dave Denkenberger: Yeah, I think he was saying that the .1% was the median, and the mean was closer to 1%.
Robert Wiblin: Oh, interesting.
Dave Denkenberger: So it would be fairly close. But yeah, you could certainly argue that it should be half as much or a third as much as this. It turns out that the conclusions don’t change very much.
Robert Wiblin: I suppose it’s also true that you haven’t considered any scenarios other than the nuclear war, although that probably is the biggest one.
Dave Denkenberger: It does dominate the asteroid and the super volcano, yeah.
Robert Wiblin: If you had to critique this number and say it as a lot lower, what kind of argument would you make?
Dave Denkenberger: Well, I would say the arguments you have made to basically how much should we update based on the evidence. If the evidence is no war in the last 72 years, then there’s a pretty strong update.
Robert Wiblin: I think Anders also claimed in my podcast with him that the fact that we’ve had so many near misses and haven’t had use of nuclear weapons could actually point in the other direction that this final step from a false alarm to actually using nuclear weapons is in fact extremely implausible. Because otherwise, we wouldn’t have as many, we wouldn’t be able to see as many near misses, because we would be dead if it was a near miss. Have you heard that line of reasoning?
Dave Denkenberger: I did listen to that good podcast, and that was an interesting point. But I think the counterpoint is we could’ve also been lucky.
Robert Wiblin: Yeah, so you just have to weigh those out. I’ve got to admit; I did find that a bit counterintuitive. Maybe you and Anders should talk and see how much you should update that number. I suppose another argument for being lower is maybe just that people find this implausible on its face. So they have this outside view, or that they do a common sense check and say, “Yeah, I just don’t think that the risk of a nuclear war each century is 38%,” in which case I guess they should adjust it lower. But I suppose you’re saying, even if you halve it or third it or quarter it, it’s still going to look pretty cost effective.
Dave Denkenberger: Yeah, you could also, talking about the outside view, think of the Fermi paradox that many people have explained the Fermi paradox; as in most civilizations destroy themselves with nuclear weapons, or something else.
Robert Wiblin: Or something similar. Okay, so let’s move on to the next key figure, which is I guess the likelihood that a nuclear war wrecks the long-term future of humanity. What was your estimate for that, and would you like to justify it?
Dave Denkenberger: So this went one, unlike being based on a quantitative model, it was based on a survey of existential risk researchers; unfortunately, only a sample size of seven but I figure it’s better than just me coming up with the numbers. Interestingly, my number was pretty consistently close to the median that there were people much more pessimistic and much more optimistic than I was.
Dave Denkenberger: But when you run the mean, the number we’re getting is around 17% reduction in far future potential because of full-scale nuclear war which is actually in the ballpark of what was on your analysis at 80,000 Hours.
Robert Wiblin: So there people are envisioning that there’s a risk that it might, through a cascading series of bad luck, cause us to actually go extinct. Or that, yeah, we could just become morally worse because of this disaster, and less able to cooperate with one another. Are there any other key considerations that you had to think about when you were coming up with your estimate for that?
Dave Denkenberger: Well certainly, the losing civilization and not recovering, even if you don’t go extinct.
Robert Wiblin: Like technological civilization falls apart, and then we don’t become advanced again?
Dave Denkenberger: Right and I should say that there are different ways that people define civilization. I would say most people use it in terms of technological civilization; so a loss of industrial civilization, and that’s more likely to happen, but not as bad.
Dave Denkenberger: I actually focus on what’s called, what I call anthropological civilization, because it’s the way anthropologists define it. That is, basically, cooperation outside your local clan. So they say the first civilizations was when we actually developed cities and larger scale cooperation. That scenario is less likely to lose that civilization, but would be much worse. Because then you might be reduced to just a few million population, like we were before we had that civilization.
Robert Wiblin: Yeah. Given that it seems like even in an all-out nuclear war, where we didn’t have very good food alternatives, there’s a good chance that 10, 20, 30% of people would survive, maybe even more than that. Why wouldn’t we recover?
Dave Denkenberger: I personally am fairly optimistic that we would recover. There’s been some talk about why we might not recover. For instance, we don’t have the easily accessible fossil fuels anymore. A guy named Lewis Dartnell who wrote the book The Knowledge, is about if we lost civilization, how could we recover it? I think it’s a great book. I’d recommend it. He also wrote an article saying well, how might it work if we don’t have fossil fuels? I think there are still routes of recovering, but it would be more challenging.
Dave Denkenberger: Another one is potentially something that wiped out the grass family, like we talked about earlier because so many civilizations have been built on the grass family, corn or wheat. Now, the exception is potatoes, like in South America. Then you could say, well, maybe the trauma of the catastrophe would select for people who are nastier or not trusting. So maybe we wouldn’t develop that trust of people who are not related to us. I don’t know how high probability that is, but it’s I think at least possible.
Dave Denkenberger: Maybe the disaster somehow permanently affects our IQ or some other characteristic that made development of civilization not possible. I think Eliezer Yudkowsky said something like we are the least intelligent that is it possible to develop industrial civilization, as in we just were able to cross the threshold.
Dave Denkenberger: Because it’s interesting that it took us something like 100,000 years from anatomically modern humans to develop from where we are now. If we had just had the knowledge we have now, we could’ve grown at 3% population growth per year, and it would only take a few centuries, but it took us 100,000 years.
Dave Denkenberger: Like I say, I’m optimistic that we would recover, but it’s just I think not a-
Robert Wiblin: Not a sure thing.
Dave Denkenberger: Yeah.
Robert Wiblin: Yeah, I guess there’s other considerations like it would probably take hundreds of years to recover from something like this to where we are now … Or maybe not, maybe a hundred years, and in that time we would know how to make nuclear weapons or how to make weapons that are much worse than what people had historically. Plus, yeah, there would’ve been selection for people who are more violent and able to survive in a difficult world.
Robert Wiblin: Maybe we just actually can’t go that long without a serious war, and it’s going to take a long time to rebuild. So there’s a conflict where we never quite manage to get to a really good state again.
Dave Denkenberger: Yeah and I think Nick Bostrom also considered that scenario of bouncing up and down and never achieving; he might’ve called it technological maturity or something like that.
Robert Wiblin: So 17% chance of … Yeah, either I guess through extinction or loss of values that would generate values in the future doesn’t seem too far off. Especially given that we’re so uncertain about it, it doesn’t seem like we should choose a number that’s really close to zero or really close to one.
Robert Wiblin: The next one then is the likelihood of preparations by ALLFED making the difference, compare to what we’re doing now if you don’t exist. What was your estimate of that, and what was the reasoning behind it?
Dave Denkenberger: First, it’s how much progress we have made so far relative to what would’ve happened without any work. I mentioned how people could just find on the internet the work we’ve done so far, and that could help out. Of course, we need to do the analysis at the margin. How much benefit do you actually get from spending more money on it? So that’s the baseline, and that number came out to be 4%. That’s 4% reduction in the loss in the far future if that makes sense.
Robert Wiblin: From what you’ve done already?
Dave Denkenberger: Yeah. So we’re not subtracting 4% off of 17%. We’re multiplying that 17% loss by .96, 1 minus .04.
Robert Wiblin: Yeah, okay.
Dave Denkenberger: Then, as we’ve said, if we could spend $100 million and do this research and development, planning, scale up et cetera, a plan for the scale that is, then the estimate came out to a mean of 18% additional reduction in the loss of the far future from a catastrophe. [crosstalk 01:36:51]
Robert Wiblin: So you’ve gone from 17 to 16.5% so far. Then you’re hoping to push it down to more like 15 or 14%.
Dave Denkenberger: Right.
Robert Wiblin: Yeah, okay, and what was the reasoning behind those numbers?
Dave Denkenberger: Well, again, this is based on a survey, but it’s actually, compared to the paper that I wrote about the probability of feeding everyone. In that paper, it was greater percent improvements. It’s an interesting question to say well, what’s easier, saving civilization or feeding everyone? I would say it’s probably easier to save civilization. So, if anything, if you believe the paper, this number could be higher.
Robert Wiblin: Do you know what people’s justification was for giving this answer in the survey?
Dave Denkenberger: I tried to limit how much work they had to do, so it was just volunteer if they wanted to explain it.
Robert Wiblin: Okay, so thinking that you’ve reduced the risk of civilization, or a collapse in a disaster scenario by four-point-something percent, it seems like quite a lot, just given that we were talking about only a small number of people, and the world is very big. You don’t worry that this is a bit too optimistic, or we’re thinking too much about what we’re doing relative to the billions and billions of people who haven’t heard about this book or any of these studies?
Dave Denkenberger: Yeah, I think my estimates of what might have been achieved so far were more like a 1% improvement, that I think it is valuable to have done this initial research, but some of the things might not work. Even if people know about it, they might not try it. They might do something else, so there are definitely many failure modes at this point.
Robert Wiblin: Yeah, or I guess people might have done this analysis after the fact, as we talked about. Yeah, I guess it seems like the key things that people can’t do afterwards, like these experiments that have long lag times haven’t been done yet.
Dave Denkenberger: Right.
Robert Wiblin: Okay, well let’s move on to talking about the last figure, which is the costs of the preparations. How did you estimate that, and how reliable do you think it is|?
Dave Denkenberger: Well, seeing as I’m a professor, I do think about how much it costs to fund a grad student, and do an experiment and write it up. I was thinking, well, how many grad student years does it take to do some research on these key promising ideas? You’d be doing research on quite a few ideas; then you would only invest the money on developing the ideas that are more promising.
Dave Denkenberger: Then I tried to talk to the people who are growing the natural gas digesting bacteria and found that they spent something like $10 million actually commercializing it. So I said well, yeah, we’d probably want to commercialize a few more of the promising technologies.
Dave Denkenberger: Then on the planning side, I thought about on the low end it would be briefing a few key people and writing reports, but in reality, you then need to make sure it’s current. So update the plan, and when people change power, then you need to brief a new person. But still, you could do that in quite a few countries for not that many millions of dollars.
Robert Wiblin: There, the range was between something like $30 million to $200 million; that was the range of plausible things that you were doing for envisaging this amount of preparation?
Dave Denkenberger: Yes.
Robert Wiblin: So then what is the ultimate cost-effectiveness that comes out of this Guesstimate sheet?
Dave Denkenberger: Well, then we have the whole separate section on the 10% shortfalls, but similar logic; higher probability, lower impact. It turns out a lot of the preparations would be relevant for both of them, so similar costs. But then what you end up with is, you could say a dollar per reductions in the long-term potential, which is not a very intuitive number because it’s very tiny.
Dave Denkenberger: In order to make it make sense, you need to compare it to something else. For that, I used the Oxford Prioritization Project, where they developed a model, and also fused in a model that was developed by Owen Cotton-Barratt and Daniel Dewey, and doing the same basic thing. On Guesstimate, what’s the probability of something bad happening, and the cost.
Dave Denkenberger: One key difference was they were looking at the marginal cost-effectiveness of just one dollar. So in order to compare this to the cost-effectiveness with alternate foods, we actually have to convert my analysis of spending $100 million into a marginal analysis. So there are two numbers you could use.
Dave Denkenberger: One is the marginal analysis of the last one hundred millionths dollar, and that’s going to be less cost effective than the average. But if you say the marginal cost-effectiveness right now, then that is going to be higher cost-effectiveness than the average. So, basically, I’ve developed a comparison with all of these numbers.
Dave Denkenberger: What you would like to know is what’s the ratio of cost-effectiveness. It turns out that when you deal with numbers that are very uncertain, you get very unintuitive results. Alternate foods is 20 times as cost-effective than AI, but AI is twice as cost-effective as alternate foods, which doesn’t make sense. When you multiply it out, the reciprocals, you actually get one, which is really bizarre.
Dave Denkenberger: But because that’s so non-intuitive, I had shifted … And that’s what I did on the original EA forum post … I, instead, shifted to the perspective of what’s the probability that one is more cost-effective than another? So, if that is 50%, then you would say, well, they’re about equal. Now, in reality, if there’s a different spread, then you could still say the expected cost-effectiveness of one is higher than the other, even if it’s 50% probably of being more cost-effective, if that makes sense.
Dave Denkenberger: One of the surprising things was that the uncertainty in this model was actually greater for alternate foods than for AI, and I thought it was going to be the reverse. Some people responded, “Well, yeah, I would probably have made the AI uncertainly greater,” so I wouldn’t put too much stock into that.
Dave Denkenberger: But basically, the result was, depending on what marginal analysis, which dollar you’re talking about, it was between 93% confident that it’s more cost effective than AI to 60% more confident that it’s more cost effective than AI. So again, if you want to change numbers, that number is going to change. But because the uncertainties are so large, you can actually change the number by a factor of 10, and the confidence doesn’t change that much.
Dave Denkenberger: So even if you’re skeptical on quite a few of the numbers, I think what’s robust is they have similar cost-effectiveness. I would say that to people who are very convinced of the high cost-effectiveness of AI that might be skeptical of that conclusion. I would say that, overall, AI, I do believe has greater importance. I think it has a greater probability of affecting the far future. However, alternate foods are significantly more neglected and probably more tractable.
Robert Wiblin: Okay, so the logic is, yeah, even though AI seems like it’s more likely to be the thing that the future turns on, there’s dozens, maybe hundreds of people, you could say are working on that, whereas you think there’s only a couple of people working alternative food. So, actually, a person can make a contribution on the margin.
Dave Denkenberger: That’s right.
Robert Wiblin: I’m mentioning that some listeners are looking at this, and if they looked at the spreadsheet and saw just how wide the confidence intervals are on this, might think that this is kind of what people might imagine [inaudible 01:44:32] is do, is that we plug in these numbers and then spit out numbers that are extremely uncertain. And then just go with the middle case, even though it’s so uncertain that, in a sense, maybe we just haven’t learned that much at all.
Robert Wiblin: GiveWell and [Open Phil 01:44:45], I think when they’re so uncertain, or when the cost-effectiveness estimates are so uncertain, they tend to use other ways to decide what to fund, like common sense heuristics like sensible does it seem, what the experts think. Do we believe that this project is well run?
Robert Wiblin: Do you agree with that approach that when we can’t really say how cost effective something is because we don’t know the inputs that we should adopt, or at least give a lot of weight to other decision-making procedures?
Dave Denkenberger: Well, I would say that if you are honest about your uncertainty, and you actually propagate that uncertainty throughout, then … Like in the case of the present generation, I do think that we should have some confidence that alternate foods is more cost effective than the GiveWell interventions because the distributions are not really overlapping.
Dave Denkenberger: In the case of alternate foods versus AI, then they really are overlapping, so you can’t say with very much confidence. I would still say that it’s valuable to run these models about neglectedness and scope and tractability. I think it’s great that 80,000 Hours have actually put numbers on that.
Dave Denkenberger: I still think that the next step is to say it’s not just scope, neglectedness, and tractability. We can actually try to put real numbers on this, and say what would we do? How much would it cost? What is the impact? Even with the uncertainty I still think we’re getting closer to a realistic estimate.
Robert Wiblin: So if you’re taking a Bayesian approach, then you start with your prior beliefs about how plausible this project seems. Then you update based on the cost-effectiveness analysis. In as much as the cost-effectiveness analysis is very uncertain, maybe the update isn’t that much, so you give a lot of weight to what you thought already.
Robert Wiblin: It might matter a lot whether someone wants to fund this, but whether, at first blush, it seemed like a sensible project to them in the first place.
Dave Denkenberger: Yeah. I have not explicitly done Bayesian updates. I would say that where the updates can be really extreme is when you’re comparing across causes; so the idea of how much do we value the far future? Well, that is just huge, huge uncertainty, and so it’s very, very difficult to compare the mosquito bed nets and AI in the same metric.
Dave Denkenberger: So if you have as strong prior then you say, “Oh, look at all this uncertainty in AI. I’m going to have an extreme update.” But the advantage of this analysis is that we’re saying the far future is the same value, so we don’t have that huge uncertainty.
Robert Wiblin: So that cancels out.
Dave Denkenberger: Right. So I would argue that we don’t have that, we shouldn’t have that potentially extreme update. Now, still, we could say let’s look at how cost-effectiveness analyses have fared in the past. It is generally true that they get worse over time. We get a reversion to the mean, and that could apply to AI. You could argue that it applies even more strongly to alternate foods because it’s newer, though I would say that, in terms of cost-effectiveness analyses, it’s actually not much newer than AI.
Dave Denkenberger: I didn’t really see much quantitative until the last couple of years, but still, there are many few people working on it, and you could argue that reversion to the mean would occur more strongly in the case of alternate foods.
Robert Wiblin: What’s the argument that it’s more tractable? They both seem like engineering problems where probably there’s a solution, but it’s fully explored yet.
Dave Denkenberger: Not to criticize AI, but I think many people who do criticize AI say, “Well, we don’t know what the artificial intelligence is going to be like.” We can start studying the artificial intelligence that we know now, but it could be completely different when we actually develop artificial general intelligence.
Dave Denkenberger: I think Owen Cotton-Barratt has gone through some of the pros and cons of working now versus later. I think that’s a significant con of working now, that things could really change a lot. Whereas, in the case of alternate foods, we do really understand what we need to do. So I do think that some amount of money should be spent on AI now, that we can make progress and we can get talented minds to bear on the problem.
Dave Denkenberger: But if I had $3 billion to spend on far future risk mitigation, I would actually be spending, right now, potentially more on alternate foods because we know what to do. And the catastrophe could easily happen this year which, generally, people don’t think that’s possible for artificial intelligence.
Dave Denkenberger: So I do think there’s great urgency, and some people have even argued that that should be a fourth category, in addition to scope, neglectedness, and tractability, of urgency.
Robert Wiblin: Yeah, it’s how much can we delay this?
Dave Denkenberger: Yeah. Then I would stay, still, because AI is more important overall, I would still spend the majority of my budget on AI, but I would spend a lot on alternate foods now.
Robert Wiblin: You mentioned a potential fourth criteria for choosing what problems to work on is urgency, do you have an idea of like how to quantify that kind of thing and include it in the framework as a whole?
Dave Denkenberger: Yeah, so in the framework of some of the earlier papers that looked at saving lives, when you try to put a value on a life, which of course is controversial, but used large uncertainties, then you can think about what the return on investment would be or the payback time of putting this money into alternate foods. It turned out that the payback time was very short, both in the global case, in the US case. Particularly in the US, the return on investment was between 800% and 40 million percent per year. Even if you only believe the lower bound, it’s still a really high return on investment. One way to think about this is well it’s very urgent, so we should be capturing these opportunities that pay off very quickly first and then, move to other things.
Robert Wiblin: How do you get the kind of 40 million percent? Is that like what happens if there’s a nuclear war next year or something like that?
Dave Denkenberger: Technically, I work with this idea of payback that if we complete the research and then, reduce the risk, then just looking at the probability of nuclear war, which has uncertainty in it, then it’s saying that you would pay back that investment very quickly, so just in terms of expected value. In the case of the US, a particular one, the cost per life saved I think went down to something like $1 per life, whereas what we spend now in the US is millions of dollars per life. It pays back very quickly if you’re saving lives at such a low cost.
Dave Denkenberger: Related to this is kind of a speculative idea, which I did mention on the EA Forum post recently of if we can’t convince governments to do the preparedness themselves, is it possible to fund that say for instance with EAs with some agreement saying that this would then save governments a tremendous amount of money to save their citizens because at least, the richer countries would generally be able to save their citizens’ lives, but at great cost because the food would cost so much if there were not alternate foods.
Dave Denkenberger: Then, if there were a catastrophe and if it saved the government billions or even trillions of dollars, then some of that savings could come back to the EAs and then, the EAs would be able to then put this money towards whatever the most cost-effective cause is at that point. I don’t know if this would work, but-
Robert Wiblin: Seems like it would be hard to get them to make that deal, will stick to it.
Dave Denkenberger: Yes and so in terms of making the deal, at least it’s more attractive to the government than funding it themselves, but in terms of follow-through, you could say, “Well it’s still no regrets from the EA perspective because even if they don’t follow through, they’ve still saved all the lives. It’s just they wouldn’t have quite as much impact as if the government actually paid up.”
Robert Wiblin: Yeah, this is another way of demonstrating the rapid kind of return on investment that you get in principle, even if you never actually manage to collect the money.
Dave Denkenberger: Right and that’s related to the urgency that then if we think, there are other problems that could use a lot of money at a later time, say AI, then we could have a win-win by funding both.
Present generation only focus
Robert Wiblin: Is it the case that these alternative foods look pretty cost effective to work on even if you’re only concerned with the present generation and not with kind of the very long-term?
Dave Denkenberger: Yes and that’s where I’ve done these cost-effectiveness calculations for saving lives and that is focused on the present generation. I think that even though there has been the most interest among EAs in the long term future implications, I think that it does have potential to appeal to those EAs most focused on the president generation. Of course, most of the interventions for that cause area have had much more certainty. I would say there are some exceptions to that like the recent discussion of deworming on 80,000 Hours Podcast.
Robert Wiblin: You mean the episode with David Roodman, who’s worked at Open Phil?
Dave Denkenberger: That’s right and there could be other examples like funding research for vaccines is a more uncertain than your typical global poverty intervention.
What do relevant experts think?
Robert Wiblin: Okay, so I guess another heuristic that you could use, other than cost-effectiveness analysis, is to look at what do relevant experts think. The people who know a lot about fungi think that, yeah, we could feed people with fungi and this is worth funding. Have you managed to check your work with other academics, and what has the feedback been like?
Dave Denkenberger: Well, as I mentioned when I was writing the book, I tried to get at least one expert to review the different sections. I do have a fairly interdisciplinary education so that I feel like I can talk to the experts, but it’s certainly great to actually get their review.
Dave Denkenberger: The experts have generally been saying, “Well, I haven’t thought about that question, but yeah, that seems possible.” Then we would need to do some more work to actually make it more detailed.
Robert Wiblin: Have you spoken to any security people or people who work on food security specifically?
Dave Denkenberger: Yeah, I’ve been to the conference … Basically, an international conference on global food security, and talked to people there; for instance, the head of the International Food Policy Research Institute in Washington DC. It was interesting talking about with him … As I mentioned, when talking to people outside of EA, we generally focus more on the 10% shortfalls because it’s more imaginable. Some of the bigger ones sound more like science fiction.
Dave Denkenberger: He was fairly optimistic that we would be able to handle the 10% shortfall. I said, “Well, what if it were 20%? What if it were 30%?” Then he was like, “Oh yeah, we’d be in trouble, and we’d need to do something else.”
Dave Denkenberger: Also, have talked to Professor Swaminathan who lead the Green Revolution in India, and was actually rated by Time Magazine as one of the 20 most influential Asians in the 20th century. I met him at that same conference, and actually, we had one of the ALLFED team members go to his institute in India.
Dave Denkenberger: He was quite positive on it. We did some interviews with him and seeing how he basically was frontline in increasing agricultural productivity to stave off a slow term disaster of overpopulation. He definitely saw the merit of looking at what would we do in the abrupt changes.
Robert Wiblin: Yeah, obviously, it looks incredibly cost-effective on paper. I guess what are the doubts that you have? What keeps you awake at night in that, potentially, you’ve made a mistake here? Is there anything that worries you?
Dave Denkenberger: I certainly am aware of the opportunity cost within EA that if money goes to alternate foods, then it’s not going to something else. I do think though that if you look at the marginal dollar now, I think there’s a fair amount of confidence that, yeah, this is worthwhile. One advantage is, by spending money now, we can get value of information. We can see, oh, is this turning out to be promising or not? Then we can decide later whether we want to spend additional millions of dollars, and the answer might be no.
Robert Wiblin: Yeah, so there’s a lot of learning potential.
Dave Denkenberger: Yeah.
Robert Wiblin: Is it possible to raise money from somewhere where you think otherwise the money just won’t go to a very effective project?
Dave Denkenberger: We certainly are targeting outside of EA. Seeing as I’m a professor, you have to apply for research grants. We’ve been looking at the National Science Foundation, where they’re focusing on hazards. I’ve looked at US Department of Agriculture, Department of Energy. One interesting proposal is … I mentioned the connection with biofuels. Well, one way of justifying a biofuel plant now is that the risk is that oil prices go down and your plant doesn’t make any money anymore.
Dave Denkenberger: But if you could switch the biofuel plant to producing sugar that people could eat then that makes the plant more economically resilient. That actually is maybe the most promising thing that we could justify economically now, so yeah, I am pursuing funding for that.
Robert Wiblin: Yeah, that’s really interesting. Are there any other more economics or policy-based solutions for this? For example, I think the government sometimes pays reserve electricity producers to just sit around, waiting for there to be the risk of blackouts or some disturbance in the electrical grid that they can then come in and keep things running. Is there any way that we could incentivize people to provide food in a disaster, even if they’re not providing it now?
Dave Denkenberger: Yeah, that’s a good question. In fact, we ran a catastrophe planning scenario event in Gothenburg, Sweden, as an add-on to an existential risk conference. One of the participants, who’s now a board member, Robin Henson, who’ve you’ve had on the show, did suggest that, we don’t want to just make plans and not know if they’re really going to work.
Dave Denkenberger: So maybe we could have a system such that the government says, “So get ready, and then at a random time we’re going to say, ‘Press go’, and you get paid based on how much food you produce two months from that time.” Then that can test, well, can we really switch over the production to producing food. So I think that’s a great idea.
Robert Wiblin: Yeah, that’s real interesting. I guess it’s a bit expensive, or hard to get through Congress at the moment, but maybe one day, or perhaps in some country that takes resilience really seriously.
Dave Denkenberger: It’s expensive if we’re talking about producing a lot of food, like a whole plant, like a chemical factory. But if we could do this on a smaller scale, then it might not be so expensive.
Anyone else doing estimates of agricultural shortfall scenarios?
Robert Wiblin: Have like any other academics or insurance companies maybe or governments tried to kind of do estimates of the likelihood of these agricultural shortfall scenarios?
Dave Denkenberger: Yes, one example is the Lloyds of London study that looked at 10% agricultural shocks and they did not do probabilities, though the UK government study did do probabilities, but neither of those looked at alternate foods as a potential solution. We’ve also talked with organizations like the World Food Program and they are generally saying, “Well, we’re just trying to deal with catastrophes like that have happened in the past decade, like a 1% global food shortfall that happened in 2011.” It’s been hard to get them to imagine the bigger catastrophes and that’s where EAs have been much more receptive to thinking about these type of catastrophes.
Robert Wiblin: What kind of journals does this kind of stuff end up getting published in anyway?
Dave Denkenberger: Well, we’ve been published in a few different areas. There are some journals on disasters. They focus on less catastrophic disasters. There have been some journals on food. There’s Futures and I think it is important to publish in mainstream journals to get other people exposed to the ideas, but I also think that we’ve had say three special issues on existential risk recently in the last four years. I think they’ve turned out quite well, so I think there is potential to actually have a specific existential risk journal.
Robert Wiblin: Do you think that would be hard to get off the ground?
Dave Denkenberger: I think that it could start with just one or two issues a year. I think there would be enough material for that. Then, you could see what happens.
Robert Wiblin: Some people have said that it’s kind of good for us to think that a nuclear winter will be extremely bad for humanity because it makes people much more reluctant to ever use nuclear weapons and to put even more effort into making sure that there never is an accidental nuclear war or unintended nuclear war. Is it possible that by like making the prospect of nuclear winter seem much less bad by inventing all these alternative foods that in fact you’re unfortunately increasing the risk of nuclear war because people would become more complacent about the issue?
Dave Denkenberger: It’s possible. I would say that for the countries directly involved, certainly the nuclear exchange would be terrible and the alternate foods does not mitigate those direct impacts of blast and fire. I highly doubt that the decision to go to nuclear war in the heat of the moment would be influenced by whether there’s a back-up plan. Now, there is evidence that both Gorbachev and Reagan cited the nuclear winter studies as a reason to reduce nuclear stockpiles in the 80s. It is true, we’ve reduced nuclear stockpiles by about a factor of three in the last few decades. Now, some critics have said it was more a decision of reduced cost in the case of the USSR that they were becoming bankrupt. There’s some uncertainty how much the concern of nuclear winter actually led to disarmament or reduced arsenals. I think there is some possibility that alternative foods, if implemented and actually believed there was a back-up plan that that could be an excuse to say not reduce nuclear arsenals as much as they would have otherwise. I think it’s fairly low chance.
Dave Denkenberger: I think overall we would be in a much better position with a back-up plan, but I will point out that in this Guesstimate model for the impact on the far future, I do have a parameter for moral hazard. You can adjust that if you want to play with the model.
Robert Wiblin: Yeah, how does one estimate a parameter like that like how large the moral hazard issue is going to be? Seems very hard.
Dave Denkenberger: It is. Given my energy efficiency background, I use the analogy that there’s this concept called the take-back effect, where if people install more efficient light bulbs, then maybe they leave their lights on more frequently. There have been studies that have shown that you calculate how much energy savings you should get and then, you might only get 80% of that in the actual implementation. That’s one way to think about it, not necessarily directly analogous.
Robert Wiblin: Suppose you can maybe also think of it as like as an economist, I think perhaps more as an elasticity, so it’s like if you make nuclear war half as bad, then like the probability if it goes up 20%, like the willingness to engage in it goes up some amount. I guess it’s like very hard to know what that elasticity actually is. I suppose the bad case would be where if you like halve the badness of it, then the willingness to go to war like more than doubles, but that seems fairly unlikely given that I suppose most of the people involved would expect to die in that scenario regardless because they’re going to be directly in the firing line.
Dave Denkenberger: Right, I do think that is a useful framework, the elasticity framework, but, yes, to actually have it end up worse off overall I think is very unlikely.
Robert Wiblin: Yeah, how did you end up deciding to work on alternative foods rather than saying a nuclear non-proliferation or trying to reduce stockpiles? Do you think that those projects are potentially like similarly effective or more so for someone who has the right fit to work on them?
Dave Denkenberger: Well, I have actually published a paper on reducing nuclear stockpiles because I have this model for the impact of nuclear weapons. We argued that just from a selfish perspective that countries shouldn’t have more than a 100 or so nukes because even if they were fired on another country and if there were no retaliation, you still have this environmental blowback. Yeah, I certainly support efforts to reduce nuclear stockpiles. I do think in terms of my own priorities, alternative foods is just so much more neglected that it’s a higher impact opportunity at this point.
Robert Wiblin: The thing that people talk about most of the time when they’re thinking of all out nuclear war is the US versus Russia, is there like a risk of other conflicts, for example between like China and Russia or China and the US given that China is modernizing its military? I think it will probably expand its nuclear arsenal at some and like how bad would those scenarios be?
Dave Denkenberger: It certainly could be very bad if China expands its nuclear arsenal, but what I’ve realized recently is even if it doesn’t, if it stays with its a couple hundred nuclear weapons, if there were an exchange between US and China or Russia and China because China now has developed so much and urbanized so much, the thousands of nuclear weapons coming from Russia or the US could produce a tremendous amount of smoke entering the stratosphere. Then, the several hundred nuclear weapons from China would hit potentially the biggest targets in those other countries, causing more smoke. I think there is potential in that type of exchange to block the Sun. As you’ve mentioned on a recent podcast, looking at when there’s been a big transition in who is the most military powerful country in the world that past experience indicates that might be a 9 in 13 probability of happening. Now, of course, we can’t take it literally, but I think it is concerning that a scenario like this could potentially block the Sun and also a reason why the estimates of probability of blocking the Sun though as you’ve argued might be too high in the case of just US, Russia. This could be a reason why they are actually reasonable.
Robert Wiblin: Yeah, it sounds like the [inaudible 00:18:44] was it that there’s an especially large amount of combustible material in Chinese cities just because the population is enormous and they’re very concentrated.
Dave Denkenberger: Well, it’s that it has changed rapidly in the last few decades because of the development of China. You can have a lot of people, but if they’re in rural areas and they don’t have very big houses, even in the urban areas, then there’s not that much combustible material. Now, if you just look at the total size of the economy of China, at least from a purchasing power parity perspective, it’s actually larger than the US. That’s an indication that the combustible material in China might be comparable to the US now.
Robert Wiblin: People often said, “Well, not so worried about China because their nuclear arsenal is quite small,” I think it’s like 300 weapons or something, which is on the lower end. They’ve been like fairly responsible, but I suppose if Russia or the US went to war with China, then it like matters how many Russia and the US has. If they do an all-out strike on China, then you could end up with a lot of fire storms and lot of material potentially going into the upper atmosphere.
Dave Denkenberger: Right.
Low figures seeming too amazing
Robert Wiblin: Some of the figures you give for the how cheap it might be to save the life, it’s like [inaudible 00:19:51] under $100 per life might strike people as kind of amazing, especially if [inaudible 00:19:57] saving lives in the present generation, not even considering the long-term future. I wonder how much should we have kind of a prior or a preconception that just the probability of being able to save lives that cheaply should be very low, so we should be skeptical of any model or any organization that claims to be able to do that. I’m thinking here a lot of people have this sort of principle that it would be like a surprising situation to find yourself in, where one person can affect like a vast number of other people and save them at very low cost, unless you have something where like everyone has to act well simultaneously in order to save one another. It can’t be the case that kind of most people are in a situation, where they can have a very outsized effects on other agents because just there’s always going to be kind of more people on the receiving end than the saving end. Do you have any thoughts on that?
Dave Denkenberger: Yeah that’s an interesting perspective. I guess if I tried to think about my prior, before doing this work, I would probably have very large uncertainty bounds. I guess why I would consider the possibility that one person could have a really big impact might be related to the fact that I think there’s a probability that I could invent something that’s really important. Maybe that’s being too confident, but it is true that some people do invent something that’s really important and have a big impact. Even though I would say this analysis ends up with quite a bit of uncertainty, I think it’s a lot less uncertainty than I would have had going into it. That I would have had even larger bounds. Still, I would update based on my super large uncertainty, based on these numbers towards those numbers.
Robert Wiblin: Yes, we had this question of like what [inaudible 00:21:41] should we have against being able to have a huge impact is kind of interesting and probably should get a section on another episode because it kind of can’t be a definitive argument against I think if you can have a larger impact because we know that there are some individuals in the past who have managed to have really large impact through like scientific research that managed to save potentially tens of millions of lives or through like catastrophic political outcomes that managed to cause the deaths of tens of millions of people. If you were like so skeptical about that ever being possible, then probably you would simply like doubt your own eyes and say that, “This could never happen that it would be impossible for [inaudible 00:22:14] many people,” because it’s more likely that you’re hallucinating or something like that rather than that had actually happened. If you kind of took this like extremely skeptical view then, it’s very, very unlikely for one person to effect such a large number of others, but yeah there’s kind a lot of thorny like questions about well how [inaudible 00:22:32] you should have and how you should update. Probably we should go into more detail in some other episode.
Our resilience to large population losses/ Rob’s scepticism
Robert Wiblin: One thing that occurs to me is that I’m not an expert on this, but I think I recall that in various like past disasters, various societies have lost 10% of the population or more, so I think in World War II, the Soviet Union lost about 15% of its population and that society kind of basically continued afterwards, although it was like a horribly scarring experience. I think in some pandemics in kind of the Middle Ages, it was common for 10, 20, 30% of the population, at least in a region to die and those societies did kind of carry on. It makes me wonder like perhaps we’re a bit more resilient to those kinds of population losses that we might get in a nuclear war or in a global food shortfall.
Dave Denkenberger: I think it is good to look at historical examples. One thing that could make this a little different, if we’re talking about a 10% global food shortfall perhaps causing 10% of people globally to die, then locally that’s going to be more than 10%. Then, of course, some places, it’s going to be less extreme, but I think what I’m most concerned about with these 10% scenarios is the spiraling downward is the way I described it, like tensions are high. Then, maybe there could be further conflict, even leading to a full-scale nuclear war.
Robert Wiblin: Yeah, another thing that a listener wrote in to ask about was that in [inaudible 00:24:46] model of a kind of full-scale nuclear war, [inaudible 00:24:49] being one of these experts in nuclear winter, it seems like most of the southern hemisphere, yeah most of the Southern Hemisphere would have like much less severe temperature declines I guess because less of the particulate matter is making it down there, which would potentially allow for kind of ordinary agriculture to persist in Australia and New Zealand and Chile and things like that, which I guess the Southern Hemisphere is not super populated compared to the Northern Hemisphere. It suggests that we might expect tens or hundreds of millions of people to survive potentially fairly easily regardless, which potentially like narrows the scenarios in which kind of ALLFED prevents, like, yeah a major collapse of civilization.
Dave Denkenberger: Well, I think the temperature falls were less extreme generally in the tropics, though they do expect the smoke to spread globally, so that there would be significant losses of temperature in the Southern Hemisphere. The other thing is that larger continents, being farther away from the ocean, they’re going to have greater temperature drops. Yeah, it certainly would be less severe in certain areas. Now, I think it does depend on the details of how the scenario plays out. If there’s not enough food for everyone and people can move towards where there is food that is if it’s not protected, then still potentially everyone could starve. People have talked about, “Well, are Islands more isolated,” but assuming that we still have most of industry functioning and have large ships, there could be basically people trying to get to those islands and get that last bit of food.
Robert Wiblin: I see, so you’re thinking that people could kind of like invade New Zealand to some extent if New Zealand’s doing particularly well?
Dave Denkenberger: Potentially yeah.
Robert Wiblin: Hadn’t thought about that. Yeah, I still kind of feel like there’s potentially quite a narrow range of scenarios in which ALLFED makes the difference between a permanent collapse versus not because the scenario in which ALLFED makes the biggest difference is one where so many people have died that kind of humanity is at risk of not recovering or having further conflict that just puts us on a downward spiral. On the other hand, you need the damage to be not so great that like the preparations that ALLFED has engaged in, kind of the information that is disseminated or the experts that it has or whatever else are not blown away or just made redundant by later efforts that the people engage in at the time. Do you think this is kind of a factor that reduces the cost-effectiveness of the approach at least even if it’s not decisive?
Dave Denkenberger: Well, I think to clarify what you’re getting at is the long-term future implication like the losing civilization. I think even in these smaller catastrophes, if alternate foods could reduce the food price, we can reduce the number of fatalities, so that we could save lives in those scenarios. In terms of the losing civilization, I mean there are people on the pessimistic end that are worried about if we run out of oil or have another financial crisis or two degrees Celsius slow global warming. You hear people talking about, “This is going to put us back into the Stone Age. This is going to destroy civilization.” I think that’s too pessimistic, but as I said, once you start getting into the 10% food shortfalls, I think the major worry about losing civilization is not the direct impact, but what happens afterwards.
Dave Denkenberger: Then, on the extreme end, I would say even if you get full-scale nuclear war and devastation of infrastructure in target countries, still you have most industry functioning globally and even if the Sun is blocked, fossil fuels are not dependent on the Sun. I think that there definitely is opportunity to save civilization by providing a lot more food. Then, the other thing about, “Well, would this happen anyway, would people realize well we need to have food sources that are not dependent on sunlight,” I’m very concerned as I said about the immediate reaction. If it is a scenario, people realizing that most people are not going to survive, then that’s a terrible scenario to be in and could result in further conflict and chaos, whether it’s within countries or between countries.
Dave Denkenberger: I think that if people were aware ahead of time or we could get the message out very quickly in a catastrophe and people realized that if we cooperated, we could actually feed everyone, then we’re more likely to do it. Of course, it’s not guaranteed.
Robert Wiblin: I guess the case for ALLFED kind of rests on this vision of like what society would look like afterwards and I guess the risk that conflict could then be the thing that really like knocks us off at the second stage and that’s why we want to prevent. Do you think it’s worth kind of engaging in further study to try to like clarify what would the world look like and how would people be behaving in that scenario or maybe it’s just so hard to know what like a post-nuclear war world would look like that thinking about it more is just not going to really clarify things. It’s just always going to remain [inaudible 00:29:27] and we kind of have to prepare maybe for the worst.
Dave Denkenberger: I think it’s going to be hard to say anything definitive, but I think that one of the projects we’re interested in doing is estimating what each country might be able to do on its own in terms of having food storage that is not prepared, any more preparations or with alternate foods, how much food could they produce. Generally, people, if they feel like in their country, they are self-sufficient in food, they would be more willing to trade, whereas if they feel they’re not self-sufficient, then they might ban trade, which would be a bad outcome overall. I think there are things that we could say usefully with some more analysis. One of the projects is doing using geographic information systems, GIS to kind of map out what the resources are and then, potentially look at possible scenarios with different levels of severity.
Robert Wiblin: Yes you’ve mentioned that to some extent ALLFED has been limited by the amount of money that it has access to. How much money you’re looking to raise at the moment and what do you think like any extra donations or unexpected donations that you get would be put towards?
Dave Denkenberger: Yeah, we have fairly detailed plans of using around $1.5 million over the next 12 months. This may sound very large because we have not used nearly that much money so far, but it’s because we’ve developed alliances with other academics that are experts in particular areas like mushrooms, rabbits, chickens, cellulosic biofuels etc. We’ve basically set up such that if we had the funding, then we could do these projects. It would also involve recruiting undergraduate and graduate students largely at Michigan Tech and University of Alaska Fairbanks, where I am. Some of the things that we would like to do, we’ve actually just started grinding up tree leaves and extracting food from them and using some fancy equipment to analyze the toxicity. We’d like to do many more leaves to figure out which ones are the most promising. As we develop this natural gas digesting bacteria solution for the household scale, we would like to actual produce a practical kit and have how-to videos of instructing people how to use it.
Dave Denkenberger: We’d also like to do plant growth experiments, simulating nuclear winter scenarios. This is a good example of something we’d want to know ahead of time because in a catastrophe, we don’t have time to experiment with lots of plants over many months to let them mature. We’d want to figure that out ahead of time and then, know where to plant specific plants. This is a scenario, where we still have some sunlight, so we’re talking about the tropics, where it likely would not freeze. We also want to investigate financial mechanisms for funding the work, for instance, through insurance or catastrophe bonds. We have a person on the team Sahil Shah, who’s looking into that. I mentioned the GIS to map out alternate food resources and analyze cooperation scenarios from economic, political science perspectives.
Dave Denkenberger: We’d also want to do experiments on turning cellulose or leaves and such into edible sugar. I’ve mentioned how that’s already being turned into sugar, but that’s not human edible at this point, so we would need to do some experiments to figure out the processes to actually make it human edible. Then, we would be ready to say how fast could we retrofit existing industrial facilities or if that turns out to be too difficult, perhaps fast construction of new facilities.
Robert Wiblin: Yeah, are there any other doubts that donors raise and that you’d like to respond to here preemptively?
Dave Denkenberger: Well, certainly a common one, especially outside of EA is this idea that if you put a lot of effort into mitigating a catastrophe that never occurs, then would the effort be wasted? I use the example of you may pay for insurance that you never use and we don’t consider that wasted because we’re insuring something that we can’t afford. I think of this work as an insurance policy for the Earth.
Robert Wiblin: Yeah, so how can listeners donate? If they’d like to I guess potentially have an email or like phone call with you, so like to clarify any other questions that they have?
Dave Denkenberger: Yeah, it’s easy to donate our website, allfed.info. Yeah, I’d be happy to have one-on-one conversations with interested people.
Robert Wiblin: Yeah, I guess your email is on the website there as well.
Dave Denkenberger: Yes.
Hiring? Who are you looking for?
Robert Wiblin: Are you hiring and what kind of staff do you need? Yeah, what sort of listeners would you like to potentially apply to work at ALLFED?
Dave Denkenberger: Yes, we do plan on hiring. Of course, how much depends on how much money we get. But I think that there’s the ALLFED proper where we’re doing more of the awareness and media work and then there’s the actual research that might be based at a university. As you mentioned, I’m based at University of Alaska Fairbanks, and then the original co-author on the book Professor Joshua Pearce, is in Michigan Technological University in Houghton, Michigan, also in the US. And he has done a lot of experimental work with 3D printers, he does a lot of appropriate technology for less developed countries. And so we would hope to be building up teams of undergrads and grad students to advance the research.
Robert Wiblin: Yeah, in what kinds of fields?
Dave Denkenberger: A lot of engineering of course, chemical engineering, we talk about chemical plants, mechanical engineering, building systems, like we talked about making sure buildings can handle growing mushrooms and the cold and such, but really a lot of fields. I mean, economics is extremely important for estimating the cost of these and for some of the low-level studies of how much does this engineering equipment costs, that’s more engineering economy. But economists tend to be more bigger picture. But I think there is important bigger picture implications here like, what are the prices of things going to be? Well, you can’t just assume their current prices, even if you know food was going to be more expensive, or maybe the inputs, maybe electricity is going to be more expensive. So it would be great to have an economist that knows how to do general equilibrium models such that we could get a better handle of how the whole system would respond.
Robert Wiblin: How about food scientists or agricultural people?
Dave Denkenberger: Absolutely, yeah.
Robert Wiblin: So you’ll be interested in potentially taking them on as PhD students or as postdocs to work some of these various different options?
Dave Denkenberger: Right. Or funding them in their home institutions if that’s more effective.
Robert Wiblin: Are there any other roles that you’re looking to fill?
Dave Denkenberger: Yeah, so we’re also interested in volunteers. We’ve benefited greatly so far from volunteer time and we’ve put several dozen effective theses on the site Effective Thesis, which by the way, I think is an excellent idea to better utilize people’s thesis time. And other volunteers could support with media, admin, et cetera. There’s also a maker component, this movement of people making stuff themselves. We hope that we can do research and say, “Well is it feasible to grow natural gas digesting bacteria in your house and then maybe develop instructions?” But then we want to know if it actually can be done by the average person or at least not an engineering grad student. And so then it would be great to put these instructions out and have people try it out, see if it works.
Robert Wiblin: What about operations people just to help scale the organization?
Dave Denkenberger: Yes. Now we have been very fortunate among the EA community to have a great Director of Operations Sonia Cassidy, who has a lot of experience managing people and operations in other companies. But yeah, certainly if we grow rapidly, then we would need more.
Robert Wiblin: Reminds me, I don’t know how big you are now. How many staff do you have and what does your budget look like?
Dave Denkenberger: So the team is around nine people, but many of them are volunteers. So at this point, it’s two or three full time equivalent.
Robert Wiblin: Yeah, so what are some kind of alternative careers that people could pursue in science or in government or perhaps other areas like academia that are relevant to alternative foods?
Dave Denkenberger: Yeah, one I haven’t mentioned was biology of course because many of these alternate foods involve biological organisms. Certainly, government jobs could potentially influence funding for alternate foods and also preparedness planning. Climate science is relevant. I think it’s a little less neglected than the actual work on alternate foods, but there are some things that would be useful on that. Political science I think I’ve mentioned with economics as well, but thinking about how much cooperation we might get in these different scenarios.
Direct work vs. ETG for ALLFED
Robert Wiblin: Do you have any view on this issue of whether if someone wants to work on alternative foods, whether they should kind of go and work for ALLFED and related projects or whether they should do earning to give if you’re particularly short on money or like fundraising is one of the bottlenecks that you face?
Dave Denkenberger: Yeah, at this point, we do have a number of collaborators, who are willing to work on this project and have the relevant skills, but we’re mainly lacking funding at this point. Yeah, I would encourage people to consider that option if they wanted to help out with this to make money to donate.
Acknowledgements/ first donors
Robert Wiblin: It’s always kind of hard to get a new organization or project off the ground. Are there any people who’ve been particularly helpful, perhaps would be able to help people get other projects off the ground?
Dave Denkenberger: Well, yeah, I’d certainly like to acknowledge the efforts of the rest of the ALLFED team, donors, including Center for Effective Altruism through an EA grant, Adam Gleave through the EA lottery, Jacob Trefethen, Greg Colburn and Ben West and a number of other donors. We appreciate all the support we’ve received.
Robert Wiblin: Yeah, how did you manage to kind of find those, the people who were willing to give you the very first donations to get things rolling?
Dave Denkenberger: There’s been quite a bit of response to the EA forum posts that I’ve made, kind of both scoping out the effectiveness of the cause area, but then specifically looking at what ALLFED can do.
Biggest mistake for ALLFED so far
Robert Wiblin: What’s the biggest mistake you think ALLFED has made so far?
Dave Denkenberger: I would say that early on, we had this idea of having an essay contest with cash prizes for people to think of additional alternative foods or maybe contribute in some way to the idea. Our idea was to get more people interested in it and we actually contacted several hundred Agricultural departments around the world with this essay contest, but overall, we didn’t get nearly as many submissions as we’d hoped for and didn’t really turn into much follow-up work. It turned out to be a mistake in retrospect, though I am optimistic about the effective thesis that we’ve put a bunch of ALLFED related theses on.
Robert Wiblin: Cool, yeah, we’ll stick up a link to that.
Robert Wiblin: All right, let’s move on now from feeding the world and ALLFED to bigger picture career issues and other things that you’ve done. One thing that I can’t help but notice is just that it seems like you’ve been extremely creative in thinking of all of these ideas and extremely open-minded. You’ve been willing to go places that most people wouldn’t just because they think it’s too strange. How do you come up with these ideas and figure out which ones to continue pursuing?
Dave Denkenberger: Well, it’s been on a long history. Even starting in sixth grade, I started having a list of my inventions and I’ve accumulated over 1000 of them so far, and I do have a patent on the heat exchanger at this point. And my other potential route to impact is making lots of money on a patent and donating lots of that. But it certainly is much more competitive, many people are trying to do the same thing. But I think that the skills of being creative and just not accepting the status quo, not saying, “Oh, if it ain’t broke, don’t fix it.” Say, “How can we do this better?” I mean, okay, yes, we could solve the problem if we spent trillions of dollars on stored food, but that’s not going to happen. So we don’t really have a solution, we need to be able to do better and yeah, basically, just not accepting that and continually thinking that there are solutions out there, we just need to find them.
Robert Wiblin: Yeah, have you had to deal with a lot of people saying, “That’s just really weird.” Or perhaps not taking your work seriously because it’s not in the mainstream?
Dave Denkenberger: Well, certainly the first reaction is, “You want to feed everyone when the sun’s blocked – that’s impossible.” But then I say well, I try to take them through the numbers. But yeah, I’m certainly open to any other feedback on why it might not work and I might have more work to do.
Engineering approach within EA
Robert Wiblin: Do you think that the Effective Altruism community maybe doesn’t have enough engineers? I don’t know a whole lot of engineers. And it seems like you bring a particular worldview.
Dave Denkenberger: Yeah, I think it was 7% in the latest EA survey that was engineers, which is not negligible.
Robert Wiblin: Okay, it’s not so bad.
Dave Denkenberger: I think that we do need a lot more problem solving. I mean, of course, we need to be aware of the problems but that’s not enough. We need to actually solve problems. And that’s what engineers do. And I think that’s illustrated by this paper written by scientists saying like, “Oh, well, humans are going to go extinct, and that’s what’s going to happen.” And engineers say, “No, we’re going to solve that problem.”
Robert Wiblin: I guess it seems like just part of your method is to just do the maths, and then just follow it to its conclusions and accept it even if it’s kind of unintuitive. Would you say that that’s right?
Dave Denkenberger: Yeah, that’s right. That’s a common way of summing it up when when people say it’s impossible. And I say, “Well, look at the numbers and then see if you think it’s impossible.”
Robert Wiblin: Yeah. Has that ever led you astray? Perhaps where you’ve investigated something that was implausible on its face, and then look to you like it was good when he did the maths, and then it turned out actually, the common sense was right after all?
Dave Denkenberger: Oh, certainly, many of my inventions have turned out to be impossible, especially when I was younger or just impractical or invented by someone else. So there’s a huge amount of failure involved in inventing, and I guess that’s the other thing I would say that you have to be not afraid of failing, if you want to take on these really hard problems. But my attitude is, “Oh, okay, well, I found something that doesn’t work. Well, I’ll focus on the ones that might.”
Robert Wiblin: Yeah, I love that you’ve got a scrapbook of 1000 different ideas for inventions that you’ve had. What are some of the most promising ones or at least most interesting ones?
Dave Denkenberger: Well, I’d say at this point, the most promising is the heat exchanger. Actually, for my dissertation, I invented a new type of heat exchanger. And heat exchangers transfer heat between liquids and gases. And so they’re used in all sorts of applications like refrigerators and cars and power plants. And it’s a crowded field, many thousands of papers, but I thought of a way of manufacturing heat exchangers more cheaply, and so they’re what are called micro channel heat exchangers that have small channels, and they have a lot of advantages that they use less material but they’re very expensive to manufacturer. And so I found a way that we might actually be able to mass produce them. And so as a part of my dissertation found out about heat exchangers or started getting interested in them, because of an application that is solar water pasteurization.
Dave Denkenberger: So in less developed countries where you have no water treatment, millions of people, mostly children die because of water contamination. And all we need to do is just heat up the water to about 150 degrees Fahrenheit, 65 degrees Celsius to kill those germs, you actually don’t need to boil it. And so you can do this in a really simple solar cooker or solar oven. But what makes the system much better is if you use a heat exchanger to take that solar energy heated water to preheat the incoming cold water and then you don’t need nearly as much solar energy.
Dave Denkenberger: And because it’s not that high temperature, I thought, well, the way of making it really cheap would be making the heat exchanger out of plastics. So I learned about laser welding of plastics, and I saw this technique of laser welding clear plastics together. So I flew to Chicago where they did this, and with a bunch of clear plastic samples, but none of them worked. And at that point, I had exhausted my free laser welding time. So I was paying the guy $200 an hour to drive to Walmart and buy Hefty ® garbage bags, and that’s actually what worked.
Robert Wiblin: Why do they work?
Dave Denkenberger: So it wasn’t complete luck. I had done some calculations on using opaque material. It just meant that the laser was absorbed and then had to conduct downward versus the laser passing through the clear plastic and then conducting upward, but yeah, you could say I built my dissertation on Hefty ® garbage bags.
Robert Wiblin: Yeah. Do you have any time to try to commercialize that or spread the word about it?
Dave Denkenberger: Yeah. So I do actually own the patent at this point. And I’ve licensed it to a company Academic Technology Ventures that basically commercializes academic inventions and so we’re talking with companies that use heat exchangers to hopefully actually get it made. And I’ve done several prototypes, but actually get it mass produced.
Robert Wiblin: What’s the biggest impediment to scaling it up?
Dave Denkenberger: As I mentioned, it’s involves laser welding, which is expensive if you do just one at a time, but if you do a continuous process or a roll-to-roll process, then you can reduce the cost per heat exchanger, but that means a large capital investment.
Robert Wiblin: I see.
Dave Denkenberger: There are also other ways of joining it, like inkjet printing of adhesives. So it’s like your printer, but you’re printing glue and that could join the pattern, join the layers together. And it could also be used for metals as well, I just haven’t done that, that prototype. So yeah, basically, we need to find that near-term application that would actually justify the capital investment to make it at a low cost.
Robert Wiblin: Yeah. So I saw recently that you posted on the Effective Altruism forum about how you’ve gotten a paper published about how to potentially prevent or reduce the scale of super volcanic eruptions, which sounds really amazing. I really would have thought that there’s nothing much that we could do about super volcano explosions. But once again, you look at that and just see an engineering problem. If I remember correctly, you came up with something like 50 possible interventions that might be able to stop super volcanoes from blowing up?
Dave Denkenberger: Yeah, either prevent it from exploding or somehow mitigate the impact of the explosion. So yeah, this is another example that there’s just conventional wisdom that we can’t do anything about it. There have been a couple papers that have considered the possibility, but I think there were five total intervention suggested, so increased it by an order of magnitude. And it wasn’t just me actually, had a co-author who was a geologist. Unfortunately, he has died since. But yeah, the I think the easiest one to explain, and probably the most cost effective is, the way a geyser works is you have this hot water under pressure in the earth. And once you get to a certain temperature, the water starts boiling. And then you get all these bubbles, and then all of a sudden, there’s less pressure pushing down on the water, which means it boils even more rapidly. So you get this chain reaction and you get a geyser.
Dave Denkenberger: Well, the super volcano actually work similarly, though it’s not water boiling. And so my idea was, well, we could at least delay the super volcano by just putting more pressure on it, and the cheapest thing would just be piling some dirt on it. And in the paper, I only talked about technical feasibility. I didn’t do economics, but I do have some non-public economic estimates that indicate it could be cost effective, even just with the present generation considered, but it’s a lot more expensive than alternate food, and it only addresses super volcanoes while alternate food address other risks. So it really can’t compare in cost effectiveness to alternate foods. And so that’s why I focus most of my efforts on those.
Robert Wiblin: But in the longer term, it’s something that we might imagine doing once we’ve taken the low hanging fruit elsewhere?
Dave Denkenberger: Right.
Robert Wiblin: So the idea is you just find a super volcano that you worry might explode, and then just pack tons of dirt on it. And then hope that you’ve put a lid on it.
Dave Denkenberger: Yeah. Now, of course, there’s the concern that that might make the eventual eruption even stronger.
Robert Wiblin: Yeah.
Dave Denkenberger: And so I have just estimated how much you might need to do to delay it to 100 years, because I figure after 100 years, we will be hopefully much more technologically mature, and then we can figure something else out.
Robert Wiblin: Yeah, what are some of the other approaches?
Dave Denkenberger: One of the original suggestions was can’t we just generate geothermal electricity from this hot volcano and also reduce the risk? Which would be a win-win. One problem is that the actual magma chamber is fairly far down. So you have to drill a long ways if you’re going to do it quickly. Normally, we wouldn’t drill that far and then therefore, it would take a long time for the heat to conduct. Whereas I was interested in what can we do to reduce the risk in one decade? But it is possible if we have drilled that far, and we have drilled at very high temperatures, but it would still require more investigation to see how feasible that is.
Dave Denkenberger: Some other ones were, it might actually be more economical just to vent the steam, then actually generate geothermal electricity, which sounds like a waste. But if you look at Yellowstone in the US, it’s pretty far away from population centers. So yeah, you can power the people nearby. But at some point, it becomes more expensive than other ways of providing electricity. And so at that point, you might just want to vent the steam.
Robert Wiblin: Yeah, how hard is it to vent a super volcano?
Dave Denkenberger: Well, if you want to do it quickly, to actually get more safety quickly, then you got to drill down close to it. And the other problem is that you might actually incite the eruption which obviously we want to avoid. So in the paper, I tried to do some qualitative technical feasibility, but also how risky is it? And I think something like piling dirt on the surface is not very risky. I mean, it’s kind of like it happens naturally with snow, or even with glacial advances and retreats. And so that’s actually one analogy we use, it looks like glacial retreats might actually reduce the probability of super volcanoes or glacial advances. So it might provide evidence that this is already working.
Robert Wiblin: Were they anymore counterintuitive ways of preventing super volcano explosions or was it largely just put pressure on it or vent it?
Dave Denkenberger: That’s mainly it. I mean, I also put in very speculative things of what we could do during an eruption. But the eruption is so powerful that those solutions would be way more expensive. So it was kind of just like, “For completeness, write down all the possible ideas, but we should focus on the ones that are most feasible.”
Robert Wiblin: Yeah, so there’s no way of capturing the material that gets thrown out?
Dave Denkenberger: So there were a couple of different classes like, could you somehow stop an eruption in progress? Well, that would be very difficult, but maybe we could have some way of disrupting the plume. Because you have all this sulfur dioxide emitted, and if it gets all the way to the stratosphere, then you get the long-term climate impact. But if you could somehow disrupt the plume, mix it somehow or cool it down or whatever, then maybe it wouldn’t get up to the stratosphere. And then you’d still have local impacts, but you wouldn’t have that global cooling.
Spending $1M to save $1B
Robert Wiblin: Yeah. Another interesting story that you mentioned in the notes to prepare for this interview, was that you’ve been involved in a team that had done some energy efficiency analysis to help make policy and found that spending a million dollars on some things that identified could save a billion dollars worth of energy. Can you expand on that?
Dave Denkenberger: Right. So the team was called Research and Policy at this company Ecos, eventually called Ecova. And the idea was to look at areas that were neglected in energy efficiency. So random small appliances. So particular study was based on external power supplies that convert alternating current to direct current for computers. And most people thought, “Oh, well, there’s these tiny things, they don’t use very much energy, and how the heck would we regulate them anyway?” Well, what we figured out and this was actually a little before my time, but then continued doing research on other projects, when I was there, was for the case of external power supplies, we could do a regulation that applied to all the power supplies regardless of what product they were powering.
Dave Denkenberger: And they developed a test procedure to measure how energy efficient they were, and then tore them apart and found out why the efficient ones were more efficient and how much more they cost, and made the case to the California Energy Commission to make a standard, and then demonstrated that this was cost effective energy savings. And then the US Department of Energy basically adopted that standard for the whole country. And so this was, less than $1,000,000 total to do this whole project over a period of quite a few years. And then of course it’s tempting to say, “The team made these external power supplies more efficient and we’re going to save energy for decades.” But that’s not realistic because it probably would have eventually happened. So really the analysis only looked at a few years acceleration and still it saved a billion dollars worth of energy.
Robert Wiblin: That’s incredible. What’s the underlying lesson here? I guess if you can get in early in something that’s growing and work on standardization then potentially you can get extraordinary leverage?
Dave Denkenberger: Yeah, and generally, use the rule of neglectedness that maybe there are some things like foot massagers, that are neglected because they don’t use very much energy. But other things are neglected, and the energy used really does add up. So it’s worth spending some effort on it.
Robert Wiblin: Yeah. Do you know why no one else was looking at this? For example the people who make these appliances.
Dave Denkenberger: Generally the manufacturers have an incentive to make the lowest cost product, and because there wasn’t any labeling, consumers had no idea how efficient they were, there really wasn’t a market incentive to make them more efficient.
Robert Wiblin: So there could be a lot of low hanging fruit there?
Dave Denkenberger: Right.
Robert Wiblin: Are there any other interesting stories or ideas that you’ve had throughout your career that you’d like to talk about?
Dave Denkenberger: Well, I guess on that team, the other projects we’ve worked on involved TVs, computers, clothes dryers is another one that I spent a lot of my time on. Everyone just assumed that you can’t do anything about clothes dryer, but actually Europe has shown that you can instead of just having electric resistance dryers that are basically like a toaster, you can use a heat pump which is running an air conditioner in reverse, basically to provide the heat and that’s a lot more efficient. Unfortunately, it’s more expensive and slower drying. So actually, what I focused more on was putting a heat exchanger on a dryer, because you have this warm air and you might as well use that to preheat the incoming cold air.
Dave’s style of thinking
Robert Wiblin: Is it just the case that if you’re an engineer, you just notice ideas for inventions all around you all the time? Or is this a more distinctive Dave Denkenberger thing?
Dave Denkenberger: Well, I mean, I’ve taught a class that’s Engineering Design. And we generally encourage students to say, “Well, what are you dissatisfied with? Where are the problems? And can you improve on it?” I think I’ve been fortunate to, to really take that to the extreme, and always be thinking about it, and make sure I write down ideas right away. I mean, one of the things, I started using voice recognition software back in the year 2000 and one of the big reasons why I did it was because I bought a voice recorder. So when I was walking around campus, and I had an idea, I could just dictate it and then keep walking, I wouldn’t lose any time, and then I’d come back and transcribe it on the computer.
Robert Wiblin: So do you use speech recognition on your computer all the time now?
Dave Denkenberger: Yeah, I do. And part of the reason was because I had some wrist issues, but even though I’ve actually finally addressed my wrist issues, still if you’re dictating more than a sentence, it’s just faster. I mean, you can go over 100 words per minute easily, even including the time of correcting it. Now, before the year 2000, I mean, value of speech recognition was laughing at the words they came up with, but after that, and you could actually talk full speed and it took into account contextual cues. It’s gotten 99% accurate.
Robert Wiblin: Interesting. Yeah, I found voice recognition of my phone really terrible. But do I just have a strange voice? So maybe it hasn’t adapted to how I sound?
Dave Denkenberger: Yeah, but the big advantage of on the computer is that you actually tell it what accent you have, and you train it for about 10 minutes. So it’s already customized. And then when it makes a mistake, you can correct it. You use some big vocabulary word that it’s saying, “No, an average person’s not going to say this.” Well, but when you train it, then it gets it right the next time. So it’s much more powerful than on the phone.
Robert Wiblin: Right. Interesting. So do you want to mention what software you’re using in case someone wants to give that a go?
Dave Denkenberger: Yeah, I use Dragon Naturally Speaking.
Robert Wiblin: And do you have problems with punctuation and putting in unusual marks and so on? Does that slow you down?
Dave Denkenberger: I do dictate punctuation, and I’m proud to say that in normal conversation I have never said period at the end of a sentence.
Robert Wiblin: Well done. So yeah, are there any advice you’d like to give to listeners about how to come up with ideas or just other advice on how to have more impact with a career that you’ve learned?
Dave Denkenberger: Some people will say, “Think about an idea when you’re going to sleep and then maybe you’ll come up with a solution.” I haven’t really found that so much. But if it works for you, that’s fine. General advice, well, I’ve been interested in effective volunteering, I think that you can have a big impact that way, that there are a lot of tasks that can be done by volunteers and maybe alternate foods is easier than other things. That there are a lot of people with transferable skills, you don’t have to learn a ton about AI, for instance, to be able to contribute. And thinking about volunteering, I think that there’s a lower barrier.
Dave Denkenberger: The’re many more people that volunteer 10% of their time or their free time, than people who give 10% of their money, and yet if you do it well, you could have similar impact. And another thing I’ve thought about which I would recommend 80,000 Hours to pursue is that there are some careers that there’s a lot of downtime that you can, I don’t know, whatever, be a security guard or something. But you could be doing effective volunteering, and still fulfill the obligations of your career. So if you think about it, if you had a 40 hour per week job nominally, but were able to volunteer for something else, 20 hours a week, that would be like 50% of your time, that could be a huge impact.
Robert Wiblin: Do you think that there are other ways to reduce existential risk that you can just do with engineering approaches? I think we haven’t recommended that that many people go into engineering, because we mostly haven’t thought that the long-term future is going to turn on engineering questions. But maybe we’re wrong about that.
Dave Denkenberger: I definitely am looking at that, because I think it is neglected. And I would like to learn more about other existential risks and see whether that might be a good strategy. But yeah, it would be great to have more people doing the same thing.
Robert Wiblin: So what was your path to working on this cause of food security? Or what kind of things did you work on before and why did you eventually shift?
Dave Denkenberger: Well, in high school, I learned about global poverty, and how much could be done, $50 could provide a year’s worth of elementary education in Haiti. And at that point, I knew that I didn’t want to have a lot of luxuries, and I wanted to donate a lot of money, and also to do high impact research to have direct impact. And so I focused quite a bit on global poverty, actually overlapped with Peter Singer when I was at Princeton, gave a sermon at my Unitarian Universalist Church on donating to global poverty, and then started working on animal welfare issues around 2007 or so. And by 2010, I was convinced by the importance of existential risk. But interestingly, it wasn’t then because of the long-term future, because I had had some economics training, and many economists discount the long-term future.
Dave Denkenberger: But I found that it could be cost effective even with the present generation and it could be of overwhelming importance if you think about the number of computer consciousnesses that we could produce even very quickly, like in 100 years. So then I moved towards the existential risk area.
Robert Wiblin: It’s extremely admirable that you’ve done this analysis and figured out that ALLFED is very cost effective, and that calls you to just start giving 50% of your income to fund ALLFED yourself. How have you found that? How do people react to that? And have you found that that’s been a burden? Or do you find it really motivating to be funding your own project?
Dave Denkenberger: When I was doing the cost effective analysis, at first I thought, “Yeah, this is cost effective, that’s great.” But then when I did the number of how many expected [present] lives could be saved for every day accelerated, that we actually get prepared and it was 100 to 40,000 lives, I said, “I have to actually donate now, this is the cause that I’ve been looking for.” And so that has enabled me, I mean, I can do a lot of the research but by funding ALLFED I’m able to pay people with other skills that can advance the overall effort much faster than if I were just trying to do it myself.
Robert Wiblin: Well, yeah. Thanks so much for all of the work that you’re doing to improve the world, and I really appreciate the creativity that you’ve brought to the table. Not so many people are willing to explore these unusual ways of improving the world.
Dave Denkenberger: Well, thank you, and thank you for your work. I’m a fan of the podcast. So it’s been great to be a guest.
Robert Wiblin: We both appreciate one another’s work. Not surprising, perhaps. My guest today has been Dave Denkenberger. Thanks for coming on the podcast Dave.
Dave Denkenberger: Thanks Rob.
Robert Wiblin: Dave wanted me to point out two things he got wrong in this episode – apparently the frequency of supervolcanos is every 10,000-100,000 years, not 100,000-1,000,000 years. On the fly Dave incorrectly converted 40 gallons to 120L instead of 150L.
Also interestingly, we looked at the most recent Effective Altruism Survey which indicates that 6% of respondents had an engineering background, so they actually probably aren’t underrepresented.
This episode is coming out around Christmas – a moment that a lot of people decide to make donations in the United States before the new tax year. 80,000 Hours has an article on the best way to decide where to give in order to have the biggest impact, which I’ll link to in the show notes.
I haven’t looked into ALLFED enough to either recommend or discourage a donation, but I’ll link to a writer on the Effective Altruism Forum who investigated it in some detail two months ago and wrote up what they concluded. If you decide you’d like to give or learn more, the site is ALLFED.info.
The 80,000 Hours Podcast is produced by Keiran Harris.
Thanks for joining, talk to you in a week or two.