Cross-posted from my blog. I question if AMF is the most cost-effective way to do good on a range of views of population ethics and the badness of death.
Word count: 1,838
In advance of the Christmas charity season, GiveWell has announced that the Against Malaria Foundation (‘AMF’) is once again their top pick for the world’s most effective charity. The explanation for this is quite straight forward: AMF provides bednets that stop very young children (mostly under-5’s) dying prematurely from malaria. GiveWell estimate, although this is not to be taken too seriously, $3,500 to AMF saves a child’s life. More specifically, that $3,500 buys 35 ‘QALYs’ (Quality-Adjusted Life Years), which is a more technical way of saying it creates 35 years of healthy life for the beneficiary. At first glance, AMF is highly cost-effective and does a lot of good. In many ways, the focus amongst effective altruists on saving lives smoothly follows from Peter Singer’s exhortation in his essay Famine, Affluence, and Morality that no decent person would walk past a drowning child. Many effective altruists (‘EAs’) enthusiastically encourage others to donate to AMF rather than their previously preferred charities, and many EA organisations, such as meta-charity Giving What We Can, use it as their leading example of how to do good better.
I believe this enthusiasm is mistaken. I doubt many effective altruists, whatever their views are on population ethics, should believe AMF is the most cost-effective way to do good. I’ll outline five different ways you can think about the value of creating and ending life. Together, they seem to exhaust the relevant possibilities. I’ve shown these in the 2 by 2 box below. I explain how much good AMF does according to each view. Then I suggest that, for each view, there are alternatives to AMF you should be tempted by instead. I’ve outlined my argument in skeleton form below. I don’t conduct a cost-effectiveness analysis of all the alternatives to AMF; that is outside the scope of this essay. So my suggestions should only be taken as suggestions. There are potentially other views in population ethics or other important considerations I’ve missed, so I would urge others to check my conclusions.
A more general worry is that effective altruists focus too much on saving lives rather than improving lives. This is understandable. ‘Saving lives’ is more motivating than ‘improving lives’. However, it’s obviously impossible to ‘save a life.’ Really, we should really talk about ‘delaying deaths’ or ‘lengthening lives’. And I take it that, if presented with a choice between extending one person’s life by an hour, or curing another from years of chronic pain, we would all choose the latter. Even if AMF is the best way to extend lives, it may not be the best way to do good. I would encourage other effective altruists to think hard about how saving lives should be compared to improving lives.
(Table 1. The value of creation and death, focussed solely on the individual who is created or dies.)
Death can be good/badDeath can’t be good/badCreation can be good/bad1. Total utilitarianism5. Reverse asymmetric viewCreation can’t be good/bad 2. Deprivation view3. Time-relative interest account4. Epicureanism
The five types of views you could take:
1. Total utilitarianism
First, you could be a total utilitarian and believe the best outcome is the one with the greatest total happiness. You can’t say AMF causes more years of happy life in the world because it probably doesn’t. Because AMF saves lives it causes families to have fewer children – parents no longer feel they need to have as many children to ensure some make it to adulthood. By GiveWell’s own estimates, the effect of AMF is that it leaves total population numbers largely unchanged. This is the ‘replacement problem’ for total utilitarians because, in these replacement cases, they can’t say there’s much (or any) value in saving lives apart from the effects of bereavement on the parents. Total utilitarians should think AMF does no or little good and should probably support efforts to reduce existential risk (‘X-risk’) instead, but lots of them already think this.
I should note it's not particularly important what the exact replacement ratio is. If it turns out AMF causes parents to have 0.5 fewer children for every 1 life it saves, the total utilitarian should still halve AMF's effectiveness.
2. The deprivation account
The second and third options are person-affecting views. These depart from total utilitarianism by claiming that whilst there’s nothing good about creating people, it is important to save the lives of people who do already exist. The basic idea is that someone needs to exist first before you can say death harms them. There appear to be to (only) two ways to count the badness of death on these views.
There’s the deprivation account, which holds that the badness of death is equivalent the number happy years of life that person is deprived of. This is the view on which AMF appears to be most effective. However, it seems somewhat implausible. It implies aborting a 1-minute old foetus is a bigger loss to that foetus than death would be for a 20-year old (if we just consider the loss for the person that dies).
If you are seriously committed to the deprivation view you should probably think it’s better to fund research into life-extension rather than save lives by conventional methods such as AMF (credit to John Halstead for this point). Nick Bostrom has suggested the chance of developing radical life-extension in the next 100 years is at least 1%, so this isn’t an option that can be discounted of our hand. Some quick maths: suppose a breakthrough allowed 8 billion people to live just one year longer on average. That would appear to be as good as extending the lives of 228 million people by 35 years, or roughly equivalent to giving $800bn to AMF (assuming they kept their effectiveness as they scaled).
(Note: Michael Dickens has written about this on the EA forum saying GiveWell take a deprivationist view. I’m not sure the view is inconsistent in the way he says it is, but that’s not really relevant here anyway)
3. The time-relative interest account
The alternative to the deprivation view is Jeff McMahan’s time-relative interest account (TRIA). It holds the badness of death depends, roughly, on the extent to which it frustrates the person’s interests in continuing to live. This captures the intuition many people have that it’s much more important to save a 20-year old than a 1-minute old foetus because, in essence, that 1-minute old foetus hasn’t developed enough to miss out on life. Advocates of TRIA should think the good AMF does is much less than the number of life years it saves as AMF mostly benefits young children, the very people the view discounts. On TRIA, the greatest loss is when someone dies in their 20s: their brain and interests have developed, but they still have many years left to live.
It’s somewhat tricky to say, if this is the view you hold, whether AMF should be your top pick. You need to reduce Givewell’s estimate of AMF’s effectiveness by however much you discount child deaths compared to adult ones. Assuming the average child AMF saves is 2.5 years old (I’m guessing), does that make AMF ten times less effective than if they were saving 20-year-olds? Half? Fifty times? All the view says is there is some discount for very young deaths. Those who hold the view will need to work out what that discount is for themselves.
It also depends on how sad you think it is for others (i.e. parents) when someone dies. Givewell’s estimates don’t include the suffering the living feel, just the benefit to the child. I don’t attempt to do this here but I make two points. First, that people adapt to bereavement (as they do to everything else) so it likely only has an impact for several months (see Oswald and Powdthavee 2008). Second, we should realistically (and somewhat grimly) expect losing a child to be less bad (but still bad) in a developing world context where there is a greater expectation that children will die early anyway.
Once you’ve bundled those figures together for a new cost-effectiveness estimate, you may think AMF is now less cost-effective than alternatives, such as anti-poverty programmes (i.e. Give Directly), mental health interventions (i.e. Basic Needs), other physical health treatments (i.e. SCI) or animal welfare charities. For reference, GiveWell thinks AMF is 4x as effective as GiveDirectly and 0.5x as effective as SCI. As I argue in another draft paper, anti-poverty programmes seem to do surprisingly little to increase happiness, whereas mental health interventions are probably much more effective than they look on conventional health metrics. I haven’t looked into SCI or animal welfare so I won’t comment further.
The fourth option is the Epicurean view, named after Greek philosopher Epicurus. It holds that there’s nothing good about creating someone and that death doesn’t harm anyone: once someone is dead, there is no them for anything to be bad. Obviously the process of dying can be painful. The point Epicureans make is that nothing is good or bad for you once you’re dead. On this account, the badness of death consists only in the suffering felt by the living.
For Epicureans, the value of their $3,500 donation to AMF is that it stops a family from having to grieve for a lost child. The question Epicureans need to ask themselves is whether that $3,500 could bring more happiness if spent on something else: anti-poverty, physical health, mental health or animal welfare interventions. Hence they are doing roughly the same calculation as advocates of TRIA, just not including the person that dies. Following my argument here, my guess is that if you want to increase human happiness, it’s more cost-effective to provide treatments for mental health than try to reduce child mortality in order to prevent parental grief.
5. Reverse asymmetrical view
The final option is what I called the reverse asymmetrical view. It combines the Epicurean view about death – that death is bad for no one – but also holds, like total utilitarianism, that it can somehow be good to create people. I don’t think this view has ever been discussed, but there seems to be conceptual space for it on the two by two grid above. I take it to be totally implausible – it holds creating people and then killing them is good, not bad or neutral, so would actually hold AMF is counter-productive – and won’t say anything more about it. I only included it for completeness.
I’ll list and respond to objections as I receive them. So far I’ve had two objections when I’ve discussed this with other people.
The first objection is that I’m suggesting we shouldn’t save lives when we can easily do so. I think this is a misunderstanding. I buy Peter Singer’s argument that, if we benefit someone else greatly at minimal cost to ourselves, we should do so. This argument starts from the point we’ve already decided we want to benefit other people and no we’re trying to work out how important saving lives is compared to other things we could do. So I’m not saying, faced with a choice between saving a life and doing nothing, we should be indifferent. Rather, I’m saying, faced with a choice between saving lives or doing something else, we should do the one that is better. It’s not clear that’s always saving lives.
The second objection is that there is some sort of deontological requirement to save lives that there isn’t about improving lives. As someone who struggles to think in deontological terms, I’m not sure how this is supposed to work. Given that giving to AMF is a charitable act, I’d have thought it was supererogatory (i.e. beyond the call of duty) which means the main concern should be how much it benefits other people. Whether giving to AMF does benefit others depends on the view you take on death, which I’ve discussed above.
AMF doesn’t seem to be obviously the best choice on any of the mainstream views of population ethics, and is definitely sub-optimal on others. This suggests, unless GiveWell wish to take a single stance of population ethics and defend it, it’s somewhat odd to call it the world’s best charity. My thought is that those initially tempted by AMF should (re)consider their own views. I wonder if it’s time for EA organisations to be a lot clearer and cleaner about the philosophical assumptions their recommendations rest on.