Leif Wenar's criticisms of effective altruism

Last week, Leif Wenar, a professor of philosophy at Stanford University, wrote a critique of effective altruism in Wired magazine. One of his complaints is that the charity evaluator websites, such as GiveWell, which sit at the heart of the movement’s efforts to encourage the world’s wealthy to donate a substantial proportion of their money and to donate it to those organizations that the available evidence suggests will do most good in expectation, do not explicitly advertise the unintended bad consequences of the activities undertaken by the charities they endorse; and, surely worse, they don’t include those consequences in their calculations of the charities’ benefits. Wenar writes that the bed nets treated with insecticide that are distributed by the Against Malaria Foundation have been used as fishing nets, resulting in over-fishing and a depletion of food supplies; and bandits have killed people who are guarding the money used by the charity New Initiatives to fund its conditional cash transfer programme. And yet, he claims, you will not find this information on the front page of the charity evaluator websites, and indeed these harms are sometimes not included in the calculations the evaluator carries out to support its list of recommendations.

What conclusion are we to draw from this? As far as I can tell, Wenar doesn’t think that, after taking these bad consequences into account, the charity evaluators would end up changing their recommendations. I don’t think he believes that these negative consequences, either those that have already happened or those we might reasonably anticipate in the future, would outweigh the good consequences of these charities’ activities. And so the natural conclusion is that it GiveWell should add more information about negative consequences to the upfront summaries they provide about each charity and their evaluation of it—as Wenar notes himself, they do in fact report these harms in their more detailed reports—and they should ensure that they always include these negative consequences in their calculations. I think Wenar anticipates that their rankings would remain unchanged, but they would now be more robustly justified and their justification would be more transparent. It’s hard to tell from Wenar’s article how often there are such omissions on the GiveWell site, but it seems true that ensuring this sort of robustness and transparency is a good thing to do—just as listing the side effects of a drug and how likely you are to experience them is a good thing for a pharmaceutical company to do.

And yet Wenar wishes to go further than this. For him, there are (at least) two problems with effective altruism that are linked to GiveWell’s apparent lack of transparency about negative effects that follow from the activities of the charities they endorse—perhaps he thinks they cause such a lack of transparency; it’s hard to tell. First, the people donating the money have no accountability to the communities affected by their donations; second, by donating money to these charities, these people are not transferring their power to the people affected, but rather choosing on their behalf what will be done for them—better, Wenar thinks, to transfer your power; and, perhaps more strongly, only morally permissible to do good in ways that transfer power.

As an instance of how instead you should donate your money, he gives the example of a friend who travels regularly to Indonesia to go surfing. That friend has met people in the local area where he surfs who are living in poverty. After talking with one of the local leaders, and hearing from him that part of what the community needs are water tanks and toilets, he has helped to plan and fund this project. Since he travels there a lot, he has accountability to those people; and since he listened to what one of the community leaders said they needed, rather than looking at their situation and deciding for himself, he is transferring his power.

A few thoughts about Wenar’s concerns about effective altruism and about his proposed alternative, beginning with the latter:

One problem is that the sort of thing his surfer friend is doing is unavailable to lots of people who nonetheless have the means to give money to charities via online donations of the sort that GiveWell recommends. The surfer has the funds and the time to travel to the same place again and again. That is a possibility only for very few, and far fewer than have the capacity to donate some of their wealth. Indeed, one reason we have charities is precisely to provide a means by which people can transfer their wealth to those who need it without having to do the travel, administration, research, networking, due diligence, anti-corruption protection, and capacity-building involved in doing the transfer themselves, since such work might be well beyond their capacities; and even if it is within their capacities, doing it would dramatically reduce the amount of money they are able to donate. Another reason we have charities is so that the wealth of many people can be pooled together to support larger scale actions where they’re needed and where economies of scale can be exploited. None of that can be done by the individualistic model Wenar proposes.

What’s more, if each person who wishes to donate to do good were to do as Wenar proposes, it would involve a vast amount of air travel of the sort that the globally wealthy should probably be trying to reduce. Perhaps Wenar will say that wealthy would-be altruists should be engaging with communities closer to home, not those it takes a long-haul flight to reach. But to do that would be to abandon some of the people suffering most. Most of the world’s wealth is geographically concentrated far from most of the world’s poverty. To encourage a sort of localism about altruism is to entrench inequality and abandon those with the greatest need.

Another problem is that there is no reason to think that the project instigated by Wenar’s friend is any less likely to have negative consequences than the activities undertaken by GiveWell’s favourite charities. Someone in the community this surfer visits must look after the donated money, and that will make them just as vulnerable to bandits as those who guard the money used by New Incentives; perhaps the outflow from the toilets that are built will rupture or overflow after an extreme weather event and cause severe illness in the community much worse than whatever was present before. That his friend is accountable to the people he is working with gives no security against these. Indeed, one of the negative consequences Wenar explicitly discusses when he talks of GiveWell charities is that the activities many of them do let governments in the affected countries off the hook, no longer needing to provide certain services they should really be providing for their citizens. But surely exactly this will happen when the water tanks and toilets are built in this Indonesian community: public amenities that the government of the country should be providing are now funded by external donors.

Going further, it is almost certainly more likely that this surfer’s project is more likely to have these negative consequences precisely because he has none of the expertise that charity workers and the aid community have amassed over many years concerning how to avoid such problems. As the examples Wenar gives show, they are not immune to these problems, but they do have greater knowledge. Indeed, my own sense is that a significant concern about effective altruism as it is currently practised is precisely the opposite from the one that Wenar identifies: effective altruists need to engage more with the expertise of the charity and aid workers around the globe, taking advantage of the huge store of practical knowledge they’ve built up over the years.

So I think that Wenar’s alternative to effective altruism is neither viable nor desirable nor indeed any improvement on effective altruism even along the dimensions he himself identifies. But is there nonetheless a problem with the approach of effective altruism? Wenar’s article seems to centre on two criticisms: first, while the activities of the charities GiveWell recommends have good consequences, they also have bad ones; second, there is uncertainty about which they will have and indeed in the past this has led charity evaluators to stop endorsing certain charities.

That something has negative consequences as well as positive ones is no reason not to do it. Allowing cars to drive around our residential neighbourhoods will, inevitably, lead to deaths in road traffic accidents, and yet people endorse this practice. Mass vaccination programmes will inevitably lead to harms, possibly deaths, for those few with previously unknown severe reactions to ingredients in the vaccines, and yet many (tragically, not always enough) support them.¹ Similarly for prescribing certain antibiotics and other medicines. Spending money on the arts that could have been spent on nurses’ salaries or medical equipment will inevitably lead to deaths that could have been prevented. We are accustomed to weighing negative and positive consequences and deciding whether, on balance, we should still do something; and we don’t consider the possibility of negative consequences an automatic veto, not least because there is no action that does not open the door to some negative consequence. But, more importantly, in many of the cases Wenar considers, the choice is not between (i) making no donation and nobody suffering and (ii) making a donation and someone possibly suffering. There is no choice here in which no-one suffers. People are currently suffering terribly in ways that are preventable.

Perhaps someone will say that at least by not donating they will not have actively caused any suffering; they will simply have let great suffering continue rather than reducing much of it while also creating some much smaller amount. But I would point again to the cases of vaccination programmes and note that we would never reason in this way in this case. We might simply let a virus spread among our community—at least then we wouldn’t have actively caused any of the great suffering and death that will result; or we might implement a mass vaccination programme, prevent much of that suffering, but also, predictably cause some. We usually choose the latter, and I hope we continue to do so.

And similarly, that there is uncertainty about the consequences of an action is also no reason not to do it. There is uncertainty about the consequences of all of our actions, and we have ways to deal with that: we weigh how likely different outcomes would be against how good they would be and come to a balanced view of whether we should do them; it is exactly these ways that decision theory tries to regiment. And again, we do this all the time even when there is a probability of negative consequences. We devise a new scheme that we hope will reduce poverty in the communities in our country. We gather evidence about it, look to similar programmes elsewhere, perhaps run small-scale trials to begin with, and then we scale it up. But things might still go wrong. After a while, we might learn enough that we shutter the scheme. Or we devise a new medical treatment for a particular condition. We gather evidence about it in trials, again start small and scale it up. But again things might go wrong. All we can do is make decisions based on the best evidence we have at the time. And, like all of science, it is fallible and we can make mistakes because our initial evidence was misleading. This can happen too with evidence about how much a given charities’ activities reduce suffering and help people. So charity evaluators sometimes endorse a charity for some period and then stop endorsing it. But what is the alternative? Wenar does describe a case in which GiveWell recommended something on the basis of weak evidence and ended up retracting its endorsement. I don’t know the details of that case, but if Wenar’s description is accurate, it’s bad. But the solution is not to give up the ideal of evidence-based decisions about how to transfer your wealth, nor to move to the surfer model of altruism that Wenar proposes. It’s to make sure you always use good quality evidence.

Let me finish with a quotation from the article:

My thought is that what [effective altruists] need first is not so much a philosophy as a good look into the eyes of the people who might be affected by their decisions. That’s what changed my path in Bali—a sense of the reality of others, whose lives are just as valuable as our own.

This is a strange criticism to hear. The usual complaint is that effective altruism is too impartial. Its insistence that everyone’s life is equally valuable—a key premise in Singer’s drowning child argument that is typically taken to motivate much of the movement—and its belief that this creates equal obligations to all people to alleviate their suffering as best you can are often taken to ignore what critics take to be our greater obligations to those with whom we have interpersonal relationships—this is a theme of this recent book, which I reviewed here. What I think Wenar means here is that, if you were to look in the eye someone who might be badly harmed by an inadvertent negative consequence of your charitable giving, you would conclude you should no longer give. And yet most would resist the analogous reasoning about vaccination campaigns. I can look a person in the eye and know it’s possible this vaccination campaign will harm them (because they will have a severe reaction to ingredients in the vaccine) and yet still endorse the campaign because I know the enormous harm and suffering that will result without the campaign. Indeed, when I look such a person in the eye, I must also remember that they are so much more likely to be harmed if the vaccinations do not go ahead. It is unclear why things should be different for charitable giving.

Perhaps Wenar will say that the accurate comparison is not with the usual sort of mass vaccination drive such as the COVID vaccination programme in the UK during 2021-22, but rather with a forced vaccination campaign, and indeed with a campaign in which those doing the forcing are now and historically have always been in a position of much greater power than those being forced. Part of his concern is that effective altruism encourages people to give to charities that do good to people living in poverty and suffering from preventable illness; they do not do good with such people; they do not work collaboratively with those people to figure out the best use to which the wealth that is being transferred might be used. I think I am more moved by this concern than others who are broadly sympathetic to the moral claim at the heart of effective altruism (even if I am rather sceptical of the apparatus built up to implement it). But it isn’t at all obvious that the charities that the evaluators recommend are often like this. Wenar gives an example of children being given anti-parasitic medication at school without the consent of their parents—though this seems to me a bad tactic even by the purely welfarist standards that effective altruists tend to use, since a parent so deceived once will rightly be sceptical of any future interventions, which in turn will reduce the possibilities of further alleviating suffering. But many other charities the evaluators recommend do not restrict anyone’s autonomy. GiveDirectly provides unconditional cash transfers to families living in poverty; the Against Malaria Foundation provides nets, but doesn’t police their use—indeed Wenar’s concern about it, namely that the nets it distributes might be used for fishing and not for protection, arise precisely because their use is not policed; New Incentives is a more difficult case, making conditional cash transfers dependent on parents vaccinating their children. This is a difficult debate. As I said, I would like to see autonomy and collaboration between charity and affected group more heavily weighted in evaluations of charities.

I think Wenar is to be commended for describing an alternative approach to the one that effective altruists take. Too often, criticisms of the movement have found fault without offering a workable alternative that recognises the extreme suffering and hardship that exists as well as the enormous relative wealth of many people in countries like the UK, the US, and Europe. But I think his alternative is worse even by the standards to which he holds altruistic action. Nonetheless, I do think we must be mindful of how charitable giving can affect the autonomy of individuals it aims to help.

1. Richard Yetter Chappell also drew the analogy with vaccination programmes somewhere on social media, but alas I can’t track it down.