HILARY GREAVES: (00:11) Okay. Hi everyone. So this talk has two aims. Firstly, I want to do a bit of first-order theorizing about cause prioritization. I want to share with you some of the thoughts that spin my head when I'm deciding which causes to donate to, which things to spend my research time on. And then secondly I'm hoping this talk will serve as an example of the ways in which apparently abstract theoretical ideas from moral philosophy might have direct real-world practical relevance. So I'm going to be talking a little bit about how one of the areas that I do abstract research and namely population ethics might end up radically affecting where you think you should be donating your philanthropic dollars.

(00:50) Okay. So here's a brief outline. I'm going to be talking about this issue of global prioritization and in particular what I see as a potential tension between two things that are very commonly thought in this area. So lots of people think that reducing child mortality is one of the best things we can try and do. Lots of people think that expanding family planning services is one of the best things we can try and do. And I want to flag and start to voice a worry you might have that thinking these two things simultaneously is in deep tension with one another. Then secondly I want to say a bit about how this looks when we start to make things quantitative, when we start to take a benefit-cost approach to our analysis. Then I'll say why population ethics is relevant to this. And then finally from my discussion of population ethics I'll float the idea that a certain theory which I'll call Total Utilitarianism might be the right abstract theory of population ethics and so then I want to know how do these benefit-cost analyses look if we agree with that? What is your benefit-cost analysis look like if you decide that your views are total utilitarian.

(01:53) Okay. So the big question of global prioritization as we've seen is: how should you best… What are the best ways in which we can try and improve the world? And you have a long list to choose from because as we all know there's a long list of ways in which the world currently is non-ideal. So you might try to do something about armed conflict, you might try to do something about immigration or water sanitation or disease. We need somehow to weigh these things up against one another and as we started to see in some of the talks this morning attempting to do this of course raises a bunch of very difficult issues. These are both conceptual issues and empirical issues.

(02:27) So on the empirical side one should think and worry about the extent to which one's proposed intervention, actually as a matter of empirical fact, succeeds in achieving the aim that it set out to achieve. Those are the empirical questions.

(02:41) On the conceptual side at the same time we face the issue that these different interventions that we're trying to choose between, say improving water and sanitation conditions or eliminating certain diseases, while those things clearly both bring benefits, they bring very different kinds of benefits. So somebody asked in one of their questions this morning, we need to be thinking about how one could weigh up those two very different kinds of benefits against one another and it's not obvious how to do so.

(03:08) We also face the point that there are inevitably going to be some tough choices there. In the prioritization arena before you start giving away your money, it's not enough to note that some cause is good, you need to know how good is it and specifically you need to know how good is it in comparison with the other things that you might be spending your money on. So before you're giving your money away, you want to know not just whether this thing is a good thing to do but whether it gets the special gold star of being a top global priority.

(03:34) Okay. All of those things said and flanked however and be the conversations as they may, there is something of a consensus around at least the following two proposed interventions. Firstly, very many people think that interventions whose aim is to reduce child mortality should be top global picks when you do this global prioritization calculation. So Give Well and Giving What We Can (GWWC), for example, have consistently recommended me Against Malaria Foundation as top choices for your charitable dollars and the calculations that back up that claim look specifically at the fact that when the Against Malaria Foundation hands out its insecticide nets one of the main effects of that is reducing deaths from malaria in children under five and in the calculations it's that component of what AMF does that leads to the very high benefit-cost scores for that intervention.

(04:25) To a slightly lesser extent, but still quite popular is the idea that we can do an awful lot of good in the world by increasing the availability of family planning in the developing world. So the Bill and Melinda Gates Foundation thinks and often says this. The 2012 Copenhagen Consensus, slightly more tentatively, arrived at a similar conclusion. This line of thought is driven especially by neo-Malthusian concerns about overpopulation. So if you think there are too many people in the world, or the population growth rate especially in developing countries is too high and if you think this is true to such an extent that it's affecting the extent to which we're able to conserve natural resources or other aspects of the environment for future generations, if you think that high population growth rates are holding back economic progress in developing countries, then you might think it would be a good thing, all things considered, if we could reduce the birth rate. And so you might think that funding family planning interventions to reach some of the millions of women in the developing world who currently don't have access to modern contraception might be one of the best ways you can spend your money.

(05:29) Okay. So I worry a lot about whether these two lines of thought can be reconciled. Here's a first pass at the worry. Suppose I buy both of these lines of argument and suppose I decide to donate to both of these causes. So I give some money to the Against Malaria Foundation and I donate some money to family planning based charities. What am I… What effect am I having in the world? Well the most obvious effect I'm having when I reduce child mortality is I'm making it the case that more human life years get lived. It's easy to get into a frame of mind where that looks like a good thing. Right? Of course human life is valuable. More of it, other things being equal, would be better. But then if I also donate to a family planning intervention, what's the main thing I'm doing there? It looks like the most obvious effect I'm having is reducing the number of human life years that get lived. And then it looks as though I need to make my mind up. Right? I either think that more human life years is a good thing in which case fund mortality reduction or I think that on balance it's a bad thing, in which case fund family planning, but how could I think they're both good things simultaneously? So that's, on the naive level, the basic worry that I want to investigate in this talk.

(06:33) So in these diagrams the time goes from left to right and each bar represents the life of one individual. So in the top bar graph, I've got a little picture of a situation where I just keep all my money for myself. I don't do this EA thing. Suppose that in that situation a particular couple, those are the brownish and the red people at the top those are the parents. In this situation, the no intervention scenario, this couple has six children and let's suppose that three of these children survive to adulthood and three of them die in infancy. Now I consider various alternatives to this scenario. Right? So in alternative number one, I give my dollars to the life-saving intervention. Maybe the result of that is that all six children survive to adulthood. So that's the second picture. So now in the third scenario, I don't fund the life-saving intervention but I do fund the family planning intervention. Maybe the result of that is that instead of having six children, this couple has three children and maybe of those three two out of the three survive to adulthood. So that's the third picture. But now look at the fourth one. If I do both of those things together then maybe in that situation the couple has three children and all three survive to adulthood. Now the point is to compare the first diagram with the last one. My interventions were motivated in the first instance by wanting to increase the number of lives that got lived to adulthood and in the second situation because I was worried about neo-Malthusian concerns I wanted to decrease the instantaneous population size so that I could encourage economic progress and so forth. But actually the net difference between the first scenario and the last is neither of those things. It's just that I've eliminated some short lives. There are three children who would otherwise have been born but died young who are not born at all in the last scenario. Either way there are three people living to adulthood.

(08:15) Now you might think at the end of the day that it's somewhat a good thing to eliminate these short lives or you might not. That discussion could go either way, but the point is it's quite hard to see if that's the net effect of all the stuff I've done, how doing that could still count as a top global priority. The rhetoric of saving lives or of speaking to any neo-Malthusian concerns about overpopulation, the things that were supposed to be driving the debate, don't seem to be able to rationalize doing both of these things at once. Okay so that's the worry.

(08:47) So now let's zoom in a little bit on that worry. When we have these explicit prioritization discussions, for example, if you look at the literature that goes into the Copenhagen Consensus, this is a global discussion where you pose the question: if you had 75 billion dollars to improve the world, what would you spend it on? And we try to take a rigorous quantitative approach to this question. So we've got a fixed amount of money to spend. We want to do the most good possible. That means that we need to be thinking about how much good we could do by each proposed intervention per dollar spent and we need to be trying to quantify the amount of good we could do for each proposed intervention per dollar spent. So that is to say we need to somehow manage to resolve these difficult decisions about how to trade off different kinds of benefits against one another and come up with a benefit-cost ratio on a common scale for the different interventions. In this quantitative approach the interventions that end up being recommended is deserving of your philanthropic dollars are the ones that have the highest benefit-cost ratios.

(09:48) Okay. So this is useful for the purposes of zooming in on the worry that I've just flagged because if it's a real worry it should show up in the calculations, it should turn out to be the case that the methodology that gets used for the calculations that recommend child mortality reduction is somehow inconsistent with the methodology that gets used for the calculations that recommend family planning.

(10:10) So I looked a bit at these two types of calculations and found that this is indeed the case. There does seem to me to be a deep inconsistency when you look at the quantitative details and the rationales for how these calculations are done. If you look at the calculations in the analysis that Give Well gives for the Against Malaria Foundation, the reason that charity comes out as very cost effective, as I mentioned earlier, is that it saves a large number of… It averts a large number of deaths in children under five per dollar spent. So if you think, and insofar as you think that saving years of life is a good thing, we can come up with a benefit-cost ratio by looking at the details of this. You can count how many years of life you add per dollar that you donate to the Against Malaria Foundation and you can calculate your benefit-cost ratio. The salient thing for my purposes is that in those calculations the neo-Malthusian concerns don't appear at all. But then you might think, “Well hang on a minute. Insofar as I really think overpopulation is a serious concern, maybe my calculation should be taking this into account.” So this is a somewhat unfashionable thing to say but it seems like it's something that we need to have out in the open. If I really think there are too many people on the planet and that this is having serious deleterious consequences for the well-being of future generations, then when I save a child's life I should be somewhat torn. I should be thinking well on the one hand I've given more years of life to this one child. That's a good making feature of a thing that I've done but on the other hand I have made it the case that the instantaneous population for the next 60 years or however long it takes for this child to eventually die is higher than it otherwise would have been and if that was really a serious concern to the to the extent that you were thinking of funding family planning to avert births, then that should be something that should be traded off in our calculation and at the moment in the existing calculations, it's not there.

(12:01) What about family planning? When you look at the calculations for family planning you see that the opposite thing is the case. The calculations that generate apparently favourable looking benefit cost ratios for family planning interventions focus entirely on the effects on other people of averting a birth. So they try to calculate, for example, the extent to which they think economic progress will be furthered by having fewer births. But again if you think human life is valuable, then you might think, “Well hang on. When I avert a birth, there's also a significant cost that I might want to trade off against that benefit. I've made it the case that 65 years of life or however many it is that would otherwise have been lived don't exist at all. Maybe that's a cost that should be traded off against the economic growth benefits of family planning interventions… Maybe.

(12:48) Okay. So I expect that many people in this room at the moment are having a thought that screams, “No. No. You're missing something. There's a perfectly good justification for doing the calculations for family planning in the way they're currently done.” So the next thing I want to do is, say what I think that rationale might be and then say what I think is wrong with it. And that's going to be where we wander into population axiology.

(13:10) Okay. So the following line of thought is extremely natural. It's extremely widespread. All of us on some level have the following intuition. Here's how I could justify funding child mortality reduction and family planning at the same time. The thing that's bad about a child's death is not just that it reduces the number of life years that get lived, full stop. The point is that there's a victim, there's a person who suffers a loss of the rest of their life. Whereas in contrast if I fund family planning interventions, yes, it results in less years of life getting lived but there's no victim precisely because I averted the birth. There is no person who suffers the loss of their life as a result of my intervention and if you think that's highly salient then you might well think that it's perfectly consistent to think both of these things are ‘top picks’ simultaneously. So this is the intuition that adding an extra person to the world, other things being equal, that is to say, if you could do so in a way that doesn't affect anybody else's well-being level, is neutral. It's neither good nor bad.

(14:09) Okay. So now this is where we get into the debates in theoretical population axiology. Moral philosophers have argued, to my mind, utterly persuasively that although this intuition is extremely natural it's ultimately incoherent. This might be one place in which you come into the philosophy room with strong intuitions but rational argument leads you to revise them. The point is the following simple one. So if this intuition of neutrality is to be something of relevance to effective altruism, it's got to be an intuition about which possible worlds are better or worse than which others. The point after all is to make the world as good as possible. We're trying to climb as high as we can up this betterness ranking of possible worlds. So the intuition’s got to be about that – if it's relevant. But in order for the intuition to be about the betterness ordering of possible worlds, it would have to be the following claim: it would have to be the claim that when you add an extra person to the world, other things being equal, you generate an alternative state of affairs that is neither better nor worse than the starting one. It's equally as good.

(15:11) Why is that problematic? It's problematic because that principle now is borderline inconsistent and the way you can see this is by noting that there's more than one possible way in which you could add a given extra person to the world and those possible ways are not all equally as good as one another.

(15:27) So in this little diagram you have A is the status quo state of affairs in which you don't add an extra person. A₁ is a state of affairs in which you add an extra person, say add a well-being level of 5. A₂ is a different state of affairs in which you also add an extra person to the status quo but you do so in such a way that that extra person is massively better off. They have a well-being level of 100 instead of 5. So now the point is clearly A₂ is better than A₁. Everybody thinks that. But then it follows that they can't both be equally as good as the status quo. But the principle of neutrality said that they are both as good as the status quo because they're both a way of adding an extra person to the world without affecting any pre-existing person's well-being level.

(16:07) And so the moral of the story is this principle of neutrality has to go. Yes, it was very intuitive. Yes, if it were true it would rationalize funding child mortality reduction and family planning at the same time, but intuitive or not it can't be correct. We need a different theory.

(16:23) Here's one alternative theory that a lot of people who've thought about this and digested the point end up thinking is somewhat plausible. It goes by the name of Total Utilitarianism in the philosophical literature. So this is the view that if you want to work out how good some state of affairs is, if you want some index of a thing you're trying to maximize, you're trying to maximize the sum total of well-being, add it up across all the sentient creatures who ever live.

(16:48) Okay. If you think that of course then you have a general default bias in favor of creating extra people insofar as you can do so without reducing the well-being levels of the people who already live because assuming that extra person is going to have a life that's worth living, it's going to be one of positive well-being that adds to the sum total of well-being in the world. The key question then if you buy this line of thought is: we've got two quantities that we need to compare against one another. We need to arrive at a consistent view on which one is bigger and in particular whether one of them is massively bigger than the other.

(17:20) So the first quantity is the intrinsic value that's contained in an individual life. Okay. So I'm considering saving a child's life. Suppose that will add 60 years of healthy life to the number of healthy life years that ever get lived. I need to assign a value to that. I need to work out how much goodness is contained in those 60 years of life.

(17:36) Second thing – if I've got neo-Malthusian concerns at all I need to try to quantify those as well. I need to ask, “If I think there's a negative effect on others, of somebody living for 60 extra years because of overpopulation related concerns, how big do I think that effect is? How much damage do I think that an extra person does through the course of living 60 years on average to the rest of the people in the world?” Come up with a number for that.

(18:01) And now the point is, however you think the results of that comparison pan out, it's going to be hard to see how you want to fund both interventions at once. If you think that the intrinsic value of 60 years of life is greater or massively greater than the negative net neo-Malthusian effect on other people, then you're going to want to fund child mortality reduction but you're probably not going to want to fund family planning. If on the other hand you think the neo-Malthusians are on to something seriously important, if you think that is to say that quantity B is much bigger than quantity A, then you might want to fund family planning but then you're relatively unlikely to want to fund child mortality reduction because you're going to be worried about the negative impact of this extra child's life on the rest of the world.

(18:46) Okay. So in summary then I've reported that both child mortality reduction and family planning are both fairly frequently cited as ‘top picks’ in this project of global prioritization. I've worried that this is prima facie a puzzling thing because it looks as though the most obvious effect of the first intervention is directly opposite to the most obvious effect of the second intervention. Then I said that this isn't just a prima facie worry, it also seems to be a real one when you look into the details of the benefit cost calculations that justify the global prioritization claims as those calculations are currently done. I think the fact that benefits cost analyses are currently done in this way is explained but is not vindicated by the fact that population axiology is a deeply counterintuitive subject. A lot of us are deeply confused about it. We really can't trust our intuitions and in particular the people, often economists, who are doing the benefit-cost analyses are not people who are either trained in or who have very close connections to the theoretical research in moral philosophy that's looked in detail into matters of population axiology. We have here a typical case of an interdisciplinary divide where you've got two groups that should be talking to one another but because of disciplinary boundaries at the moment usually aren't.

(20:04) So I think what we need to fix this is more sophisticated analysis. We need to be having serious conversations like the following one. Okay, we've got these two effects the intrinsic value of a child's life and the neo-Malthusian concern. Can we please try to put these on a common scale and have a serious conversation about which of them is bigger. That kind of question only looms large as being an important one if you've bought into a total utilitarian framework but by no means at all, I shouldn't overstate the point, many people who have thought about the abstract theoretical issues of population ethics do end up concluding in favor of total utilitarianism despite some of its counterintuitive features.

(20:42) I think it's dangerous to ignore this kind of point. I mean I was somewhat worried about even raising the issue at a conference like this because one response that some people have to this kind of concern is that they find it sort of paralyzing. You might feel that the uncertainties are just so great that we should give up on this whole effective altruism thing and keep our money for ourselves. I mean that's not my conclusion but I can see why it might lead people that way.

(21:08) My conclusion is one of the ones that one of this morning's speakers was making fun of. It's that we seriously need more research here. The reason I think it's important is that there's a worry of being really quite seriously counterproductive with our philanthropic dollars if we don't have serious conversations about this kind of thing. One kind of worry is intrapersonal. Right? So I'm an individual. I want to give away my money to make the world a better place. I am worried as an individual that if I fund both of these things half of my dollars might be counteracting the other half of my dollars. But we also face the same worry as a community. It's no better really if each of us individually makes our decision about which of these quantities is the greater one. Half of us come to the conclusion that the intrinsic value of a child's life is massively greater than the neo-Malthusian damage a child does. And so those people fund life extension. The other half of us come to the conclusion that, no, the neo-Malthusians are right, so they fund birth aversion. Then as a community we're spending millions of dollars with one half of the community's resources to a large extent fighting against the other half of the community’s resources and of course that's not a situation that we want. Thanks.

[applause]